Whenever a test result is present in base test result but absent from
target test results, we have an entry in regression report looking like the
following one:
ptestresult.apr.testfile: PASSED -> None
The "None" status may be a bit confusing, so replace it with a custom
string which really states what "None" means in this context:
ptestresult.apr.testfile: PASSED -> No matching test result
(From OE-Core rev: b5f6da6e13f3484f51dfa07f6b3500aea7a21825)
Signed-off-by: Alexis Lothoré <alexis.lothore@bootlin.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add a STATUS_STRINGS dictionnary matching raw statuses to custom strings.
Whenever a regression must be reported, raw status is searched in the
custom statuses dict (key search is case insensitive). If no custom string
is found, raw status is kept and used in regression report
(From OE-Core rev: 9d22bfc9d0c4092dba1af0ee11a4c51b7b270786)
Signed-off-by: Alexis Lothoré <alexis.lothore@bootlin.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Brings following changes
* 83b858f8 fix rejection of dns responses with pointers past 512 byte offset
* a4ecaf89 dns stub resolver: increase buffer size to handle chained CNAMEs
* 40834f6c printf core: fix gratuitous integer formatting buffer size
* 3a23cd06 all printf variants: fix argument type handling for %c and %lc
* fa4a8abd fix public clone function to be safe and usable by applications
* 0c277ff1 fix broken thread list unlocking after fork
* 4653b987 mbrtowc: Fix wrong return value when n > UINT_MAX
* b90841e2 configure: replace -Os with equivalent based on -O2
(From OE-Core rev: b6abcd4f01aec24393bce68a9806d94702c0d387)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
with newer systemd, if not specified, these paths are searched on build
host /usr dir
log.do_configure.2373278:Program /usr/sbin/loadkeys found: YES (/usr/sbin/loadkeys)
log.do_configure.2373278:Program /usr/sbin/setfont found: YES (/usr/sbin/setfont)
and if build host happens to have kbd installed then it
uses those locations, this would work fine if the assumption between OE
distro and host distro matches but fail otherwise e.g. on archlinux
these binaries are in /usr/sbin but in OE they are in ${base_bindir}
This results in
qemux86-64 systemd-vconsole-setup[211]: /usr/sbin/loadkeys failed with exit status 1.
(From OE-Core rev: 94ccc7acc4a871f5bb7ab8e135e70b5519eff6ad)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Cc: Luca Boccassi <luca.boccassi@microsoft.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
- Drop dependency on gnu-efi, add dependency on pyelftools for EFI builds
- Refresh patches
- Ship new files and directories
- Use meson target to build sd-boot instead of filenames
- Change libpam recipe to set ANY_OF_DISTRO_FEATURES = "pam systemd" to let
logind pull in pam-plugin-umask
(From OE-Core rev: 95ed1fa4ff74a77deade51ad73b2f8963ff81548)
Signed-off-by: Luca Boccassi <luca.boccassi@microsoft.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
With newly added recipe_qa task, the function needs to filter out its output as well.
The issue wouldn't appear in selftest because it's triggered by buildhistory and
selftest disables that.
(From OE-Core rev: da1284c990957827ee89685921c53b64490a32b2)
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This can happen if running unpack task produces unrelated files and directories
(e.g. if recipe_qa or other tasks run); in this case it's better to stop, rather
than allow devtool to continue and error out much later with a message that has
nothing to do with where the problem originated.
The idea here was to handle tarballs that don't contain a top level directory and thus
the source tree is one level up; this basically never happens, and if it does we
should find a less brittle way to handle such tarballs.
(From OE-Core rev: b25e922b271794906b22450c7e6cc18fcab51ff8)
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Update test_python_snippet_function_reference to not require the 'func'
flag, now that we know the real function will be returned for context
functions without the flag.
(Bitbake rev: 83f41281ec3d9b4327ffc8e2312e1fb8f53cbf02)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We previously checked for the existence of the 'func' flag to determine
if we should avoid looking up in the metadata. This was done to ensure
the user gets the function for 'def' python functions rather than their
string contents. We can sidestep the metadata lookup and check our
function context directly, instead.
(Bitbake rev: 6cac1eac51efa9a54e8457f60ea1ea0e604c50b7)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This test makes sure that '${@eval()}' calls the eval builtin, even if an 'eval' variable is defined in the metadata.
(Bitbake rev: e9150447738a48f772240874b3512b08e982b19b)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This avoids the need to hardcode a list of python builtins. This also slightly changes behavior, in a case like `${@eval("3")}`, this will ensure we always call the builtin, even if the metadata has an 'eval' variable defined.
(Bitbake rev: 9976ae50677b333d646ca0fd395468bd2301d03f)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This ensures that any change to the presence of builtins in inline
python execs will be noticed.
(Bitbake rev: ee22d3d51c60db2da97422b2be1e42239b7a2324)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is a test for an issue seen long ago, to avoid regressions, where a
reference to a def'd function in the metadata would return the string
value from the metadata rather than the function in inline python.
(Bitbake rev: 9f7cb22febd557817c164e25a93f5660e9c06358)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This has been around without being properly documented since 2019 (!!!),
and is nowadays the preferred method for enforcing license restrictions,
especially since meta-gplv2 is officially obsolete.
(From yocto-docs rev: efa1c57ecec934998792b7851b4a162be92c8b23)
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
- Kirkstone is now a 4 year LTS
- Add future Scarthgap LTS
- Update document metadata: license, title
(From yocto-docs rev: 12d40f56edc755db9724b7382e30082874f29699)
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Restore the error message as it is supposed to be issued.
The literal block renders its content verbatim so a term role (`:term:`)
would appear as :term: string in there and `---` as --- string instead
of the expected hyphen.
Fixes: e319b3bf4eb6 ("manuals: add missing references to SDKMACHINE and SDK_ARCH")
Cc: Quentin Schulz <foss+yocto@0leil.net>
(From yocto-docs rev: 95158beba0aea7539024688c71b54fed5267faec)
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
`login` from busybox has explicit heuristics to handle CONFIG_MULTIUSER
unset (i.e. a kernel without non-root users, groups or capabilities),
bringing in the shadow version of `login` causes obscure failures at
runtime ("setgid: Function not implemented"), instead exclude it here so
anyone who includes it (likely accidentally) will at least get a build
time failure.
(From meta-yocto rev: be22c2923a0cdf60ab60012668382e55a3618f2b)
Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
https://wheel.readthedocs.io/en/stable/news.html
0.41.0 (2023-07-22)
* Added full support of the build tag syntax to wheel tags (you can now
set a build tag like 123mytag)
* Fixed warning on Python 3.12 about onerror deprecation. (PR by Henry
Schreiner)
* Support testing on Python 3.12 betas (PR by Ewout ter Hoeven)
(From OE-Core rev: 81e7e3b9059840fd749b5b50fc75588d7c4b7593)
Signed-off-by: Tim Orling <tim.orling@konsulko.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
https://more-itertools.readthedocs.io/en/stable/versions.html
10.0.0
* Potentially breaking changes
- Python 3.7 support was dropped, since it went EOL on 2023-06-27
- batched() no longer issues a DeprecationWarning; it is now an alias
for itertools.batched for Python 3.12+
- batched() and matmul() now yield tuples instead of lists
* New functions
- combination_with_replacement_index() (thanks to Schoyen)
- nth_combination_with_replacement() (thanks to Schoyen)
- polynomial_eval() (from the Python itertools docs)
- polynomial_derivative() (from the Python itertools docs)
- sum_of_squares() (from the Python itertools docs)
* Changes to existing functions
- seekable() now has relative_seek method (thanks to karlb)
- chunked_even() was optimized (thanks to elliotwutingfeng)
- numeric_range() was optimized (thanks to eltoder)
- duplicates_justseen(), pairwise(), partial_product(), and partition()
were updated and optimized (thanks to pochmann)
- unique_in_window() had its implementation updated (thanks to
elliotwutingfeng)
- iterate() now breaks when its func argument raises StopIteration (thanks
to jrebiffe)
* Other changes
- Some documentation and testing issues were fixed (thanks to lonnen and
XuehaiPan)
(From OE-Core rev: 9351489f3fd9a01f1e74b4f209df01aeffe55c8f)
Signed-off-by: Tim Orling <tim.orling@konsulko.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Rewrite of CVE_CHECK_IGNORE to CVE_STATUS contained copy+paste
problem changing CVE numbers.
CVE-2020-12352 -> CVE-2022-3563
CVE-2020-24490 -> CVE-2022-3637
CVE-2020-12352 is now for kernel only in NVD BD, so remove it.
CVE-2020-24490 is corrected in this commit.
(From OE-Core rev: 150f81b764ccf1abfc69bd573d1fb997a6115884)
Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This will remove 6 CVEs which were already excluded before.
(From OE-Core rev: 8fbf7ec2b98395d87fed1360e362dc1e91e6d400)
Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This issue was reported and discusses under [1] which is linked in NVD CVE report.
It was already documented that some parts or libarchive are thread safe and some not.
[2] was now merged to document that also reported function is not thread safe.
So this CVE *now* reports thread race condition for non-thread-safe function.
And as such the CVE report is now invalid.
The issue is still not closed for 2 reasons:
* better document what is and what is not thread safe
* request to public if someone could make these functions thread safe
This should however not invalidate above statment about ignoring this CVE.
[1] https://github.com/libarchive/libarchive/issues/1876
[2] https://github.com/libarchive/libarchive/pull/1875
(From OE-Core rev: 9b5b850d6a6982bb8ff14dcfbb6769b293638293)
Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For newest Qualcomm platforms the firmware for the Adreno GPU consists
of two parts: platform-independent SQE/GMU/GPMU/PFP/PM4 and
platform-specific ZAP shader, which is used during the boot process. As
the platform-independent parts can be shared between different
platforms, split the platform-specific part to the separate package.
(From OE-Core rev: bf00a042d2fa2eb4b20d8c5982926758821bf990)
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
RTL8822 is a serie of wireless modules that need firmwares to function correctly.
The linux firmware recipe does not have a package of these firmwares, and this commit add them.
(From OE-Core rev: 6459959beeb91c0b694f5f17b6587a12c6dcb087)
Signed-off-by: BELOUARGA Mohamed <m.belouarga@technologyandstrategy.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The specific set was written to ensure vim builds reproducibly; this is
however prone to silent regressions (if vim adds more usage of iconv with
different encodings). Installing the full set also matches what standard
desktop distributions do, and thus meets expectations of upstream projects in general.
(From OE-Core rev: 2685c497c14bc325500053e551d97e8cbda97252)
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This resolves two issues:
1. metapackages were depending on themselves (except -binaries which wouldn't match against 'glibc-binary').
2. for the nativesdk variant, due to a non-empty dependency list at parsing time caused by
issue 1, map_depends_variable() from meta/lib/oe/classextend.py was forcibly setting PACKAGES
to the initial parse-time value (e.g. missing the dynamically created packages). This meant that
three out of four nativesdk- metapackages were entireyly missing the dependencies on the
respective dynamic package sets.
(From OE-Core rev: a90fd3afe9184aa1870b34a826e3ba0563477d4b)
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
