There is a buildpaths QA error in the test recipe that somehow didn't show up
in earlier testing. Use a source directory and tweak the build command to
avoid embedding build paths in the generated binary.
(From OE-Core rev: 8f1cfaf2be834217cd1cf5be98f44270c36cc31a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Most of these warnings have been around and enabled for a long time. In particular,
buildpaths has been like this for two years. I'm aware some layers still have not
been able to resolve all the warnings but I believe that regardless, it is still
time to raise the bar. If the warnings don't get fixed, it is probably a sign
that nobody cares about the recipe and it should be dropped.
For anyone coming here to find out what changed and how to disable it, if
you are going to remove from ERROR_QA and add back to WARN_QA (or just ignore
the warnings), please do it with a layer specific override rather than making
it global. We have fixed these issues in core and intend to keep them all fixed.
If you globally disable the errors, it just means we get patches which end up
regressing things.
You can do things like:
ERROR_QA:remove:layer-mylayername = "buildpaths"
not that I'd recommend it.
Also note that the next version of Yocto Project Compatible will only be
available to layers which are not disabling some set of these errors.
(From OE-Core rev: b79b191cc43a45dde2adb61ea349b426cb2461d1)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
These were used by external layers and were needed before we had better
unpack instrumenation and the recent unpack directory changes. Drop
them as obsolete.
(From OE-Core rev: 8dae8778ee14caebea92acd0bebf28c5681e39f3)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Enforce the default remote name to "origin", as assumed in numerous
places.
This prevents build failures in case the system/user configuration sets
this to a different value.
(Bitbake rev: 1d7360031164f04887c792fb0b2dd86c6ccfcc23)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Update the test cases since the implementation is changed:
* test_shallow_multi_one_uri()
The a_branch and v0.0 had the same revision, and it required fetch a_branch
and remove histories of v0.0 which were conflicted, and bitbake reported:
fatal: no commits selected for shallow requests
Make a_branch and v0.0 have different revs to fix the problem.
And now the 'rev^' is not needed, so update self.assertRevCount() as well.
* test_shallow_multi_one_uri_depths()
Update self.assertRevCount(), now git only fetches the required revs.
* test_shallow_fetch_missing_revs()
The command is:
$ git fetch --shallow-exclude=v0.0 master
But master and v0.0 uses the same revision, so there is no commit to fetch.
* test_shallow_fetch_missing_revs_fails()
Two unneeded committs are not fetched now:
- rev^
- One not specified or required tag.
So update self.assertRevCount()
(Bitbake rev: 48eff9d9a660ad6b9bd8b53a7dcec600ef42b1d1)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch can make the following settings much more faster:
BB_GIT_SHALLOW = "1"
BB_GENERATE_MIRROR_TARBALLS = "1"
* The previous implementation was:
- Make a full clone for the repo from local ud.clonedir
- Use git-make-shallow to remove unneeded revs
It was very slow for recipes which have a lot of SRC_URIs, for example
vulkan-samples and docker-compose, the docker-compose can't be done after 5
hours.
$ bitbake vulkan-samples -cfetch
Before: 12 minutes
Now: 2 minutes
$ bitbake docker-compose -cfetch
Before: More than 300 minutes
Now: 15 minutes
* The patch uses git shallow fetch to fetch the repo from local
ud.clonedir:
- For BB_GIT_SHALLOW_DEPTH: git fetch --depth <depth> rev
- For BB_GIT_SHALLOW_REVS: git fetch --shallow-exclude=<revs> rev
Then the git repo will be shallow, and git-make-shallow is not needed any
more.
And git shallow fetch will download less commits than before since it doesn't
need "rev^" to parse the dependencies, the previous code always need 'rev^'.
(Bitbake rev: a5a569c075224fe41707cfa9123c442d1fda2fbf)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
DEBUG_FLAGS is no longer needed here (and changed to DEBUG_LEVELFLAG so would break).
(From meta-yocto rev: 74f68f1a91ebfa630909216ac5c25971e7b3841c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Backport patch for CVE-2021-42380.
Additionally backport clang regression fix caused by this patch.
(From OE-Core rev: 66543769ff79d81508bb703bd2fc34871a16e2c7)
Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix is only done for target. Use same code for nativesdk.
(From OE-Core rev: dc6306883cc2c7d4d98d595442e5bf4037a160c5)
Signed-off-by: Oleksiy Obitotskyy <oobitots@cisco.com>
Signed-off-by: Oleksandr Hnatiuk <ohnatiuk@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When combining an SPDX document, the package list might be empty (e.g.
a baremetal image). Handle this case instead of erroring out
(From OE-Core rev: baf4e360f6e65a5e9aff2def69d2a720f38f92b2)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
SPDX 3.0 introduces a bunch of new SPDX tasks. Instead of explicitly
enumerating them all, modify the regex to match 'create_.*spdx' which
will cover all of the SPDX 2.0 and SPDX 3.0 tasks
(From OE-Core rev: 1d007eaf0155b5ac1c90a15634f7b1473743a4bd)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Downstream tasks may want to know what image files were written so write
out a manifest in do_image_complete. The format of the manifest is the
same as the one in image.bbclass
(From OE-Core rev: 4c4f1e020533c26f90e95960745f6de90d77e64e)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Adds a class that allows recipes to opt out of generating SPDX
(From OE-Core rev: 170277da0dc34e3f797451473ebfbde82922f593)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Adds the variable overrides to set the SPDX image purpose for various
image types
(From OE-Core rev: e1a8f59462d88c92dfdb3632ca95ded094c9472a)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
do_create_spdx is a outlier in that it doesn't need the RSS to be
extended just because it depends on do_populate_sysroot. In fact, it
only depends on do_populate_sysroot so it can see the actual recipes
sysroot, and attempting to extend the sysroot can cause problems for
some recipes (e.g. if a recipe does do_populate_sysroot[noexec] = "1")
As such, explicitly exclude do_create_spdx from extending the sysroot
just because it depends on do_populate_sysroot.
(From OE-Core rev: 8b506f327838b885bfeb2cf144f43c8be68b8591)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Downstream tasks may want to know what image files were written by the
do_image family of tasks (e.g. SPDX) so have each task write out a
manifest file that describes the files it produced, then aggregate them
in do_image_complete
(From OE-Core rev: 5e55ed4c5b9d5af3c96b82805af34af1512fc3d1)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The default busybox wget https support is suboptimal, it silently ignores
checking certificate validity which isn't great for security.
Switch our defaults to disable the internal busybox tls code and the
https support using it and configure the openssl backend instead.
This this is done by spawning an openssl command, we don't need
dependencies on openssl for build. For runtime, we can assume
people would install openssl if they need/want this.
These changes put our default busybox configuration in a more secure
initial set of settings.
[YOCTO #14125]
(From OE-Core rev: 5d4ad13462f12355ff0f2bc1773ab4b1814b165a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When the optimization flags were reduced to a single option, a different
and clearly untested codepath was used where the output was unquoted. This
caused errors when testing meson in the SDK.
Use repr() in the single item case as per the multiple item case to correctly
quote the output and fix build failures.
(From OE-Core rev: 7c2b573e6cbe518f54ea5c3539fc91ed63b80ebf)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The recent ERROR_QA changes adding build_deps exposed what were warnings in thi
test as it restricts the locales, then builds recipes with locale requirements
that were not met (in bash, gawk and python3).
Increase the list of generated locales to ensure this patches the recipes
being built and hence allowing the build to succeed.
(From OE-Core rev: ebe2df8ab6e9d4087f8ddbffbc84eea6c4405738)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Recent changes to ERROR_QA meant that the build was failing in ways the test
wasn't expecting. Tweak the tests to ensure those failures don't impact the test.
(From OE-Core rev: c206a87f2e7128308e38a046b1dcd8cc43c91125)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* Since commit [a23c482cab allarch: only enable allarch when multilib is
not used], allarch recipes will also be installed into
${MLPREFIX}recipe-sysroot, so this appending is not needed.
* This appending also causes some QA errors. Eg: for lib32-php, the
recipe will use 's@${RECIPE_SYSROOT}@@g' to remove host specific path,
this appending makes not all the host specific path are matched.
(From OE-Core rev: e6e6076f1956c711814c14d76194794d950e45f8)
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Move the -pipe option out of the optimization flags and directly into
the flags variables since we always use it now.
Also move the debug prefix mapping there to match the nativesdk case
which already does this.
Fix the documentation and two recipe usages to match the change.
(From OE-Core rev: 9badf68d78d995f7d5d4cf27e045f029fc6d4044)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
eliminate-unused-debug-types is a compiler default so we no longer
need to specify this. Drop the option. Also drop an obsolete comment
about another debug flag.
(From OE-Core rev: 1d5162689fbfeddb10fa60e4150ca87f1fa2243e)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Commit 0827c29566 (openssh: allow configuration of hostkey type) broke
our setup. We make use of the 'Include /etc/ssh/sshd_config.d/*.conf'
and put a hostkeys.conf file in there, configuring the types and
locations of the sshd host keys.
With that commit, we now get an extra "HostKey
/etc/ssh/ssh_host_ecdsa_key" line in the sshd_config. And while we
could avoid that by removing all hostkey-* items from PACKAGECONFIG,
other people providing their own sshd_config via a .bbappend now have
their HostKey settings unconditionally removed by the 'sed'
invocations, regardless of PACKAGECONFIG.
To make it easier for downstream layers and BSPs to define (and
preserve) their own logic for placement and type of sshd host keys,
factor out the new logic to a separate shell function. Downstream
layers can then simply override that by an empty function and keep the
behaviour they used to have.
(From OE-Core rev: 09dd5cceecfaa2046f7ed070690b000181723fd2)
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When building an image including iptable built with the libnftnl
PACKAGECONFIG, one hits
Downloading file:.../oe-rootfs-repo/armv8a/libkmod2 * check_data_file_clashes: Package iptables wants to install file .../rootfs/etc/ethertypes
But that file is already provided by package * netbase
This used to be handled by
0003-Makefile.am-do-not-install-etc-ethertypes.patch, but that patch
got removed with the 1.8.9->1.8.10 upgrade (commit 4616ada82e70).
I think the rationale for dropping the patch was wrong; the commit log
talks about xtables.conf, which is indeed gone from upstream, but said
patch didn't change anything about xtables.conf, it did
-dist_conf_DATA = etc/ethertypes etc/xtables.conf
+dist_conf_DATA = etc/xtables.conf
However, instead of patching iptables to not install ethertypes, and
having to forward-port that patch, it is much simpler to just remove
the file in this do_install:append.
(From OE-Core rev: 3b9c8d5c26a48123d3845c93f6e04d430569e6a2)
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This currently puts a 60MB binary into the recipe sysroots in the native
case as well as the implications for the target. Since as far as I know,
we don't need/use this, stop generating it by adding a PACKAGECONFIG. It
can be enabled by anyone who does actually need to use it.
(From OE-Core rev: e42abd3b82c7bac93541b2cf87b553c1a31346cd)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
After util-linux upgrade to 2.40.1, we got the following QA issue
on musl:
WARNING: libssh2-1.11.0-r0 do_package_qa: QA Issue: libssh2-ptest
rdepends on util-linux-col, but it isn't a build dependency? [build-deps]
The issue appeared because new version util-linux has disabled
building out col on musl[1]. In short, the reason is that col in
util-linux requires something that musl does not provide.
This 'util-linux-col' dependency was added in 2021, at that time
the libssh2's version is 1.9.0. Now the version is 1.11.0 and I tested
its ptest, it does not require util-linux-col to succeed.
[1] 8886d84e25
(From OE-Core rev: 9007de7e388e269028c7c1722f0d7ee76bc5d251)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
License-Update:
===============
GPL-2.0 is changed to GPL-2.0-only in README.licensing.
The following two patches are removed as they've been there in new release:
0001-login-utils-include-libgen.h-for-basename-API.patch
mit-license.patch
Other existing patches are refreshed to avoid patch-fuzz error.
A new PACKAGECONFIG, lastlog2, is added, which serves as a switch to build Y2038
safe lastlog pam module. This is a new functionality added in 2.40.
This 2.40.1 version has the problem of unnecessary remake, causing
intermmittent errors at parallel make. The do_compile:prepend is a
workaround for this. Upstream issue and PR are:
https://github.com/util-linux/util-linux/issues/3061https://github.com/util-linux/util-linux/pull/3064
do_install is changed to do_install:append, the legacy codes are
removed. They've been there since the born of oe-core. I tested
with ccache, there's no error.
A new dependency, diffutils, is added to ptest. Because the mkswap-offset
test case needs 'cmp' from diffutils.
The tests/ts/misc/enosys test case is deleted for musl. This is because
the test case is comparing error description output and that's libc
specific. Results below can demonstrate the case.
'''
root@qemux86-64:/usr/lib/util-linux/ptest# cat /usr/lib/util-linux/ptest/tests/expected/misc/enosys-ioctl
test_enosys: ioctl r=0 errno=Success
test_enosys: ioctl r=-1 errno=Function not implemented
test_enosys: ioctl r=-1 errno=Inappropriate ioctl for device
test_enosys: ioctl r=-1 errno=Inappropriate ioctl for device
test_enosys: ioctl r=-1 errno=Function not implemented
root@qemux86-64:/usr/lib/util-linux/ptest# cat /usr/lib/util-linux/ptest/tests/output/misc/enosys-ioctl
test_enosys: ioctl r=0 errno=No error information
test_enosys: ioctl r=-1 errno=Function not implemented
test_enosys: ioctl r=-1 errno=Not a tty
test_enosys: ioctl r=-1 errno=Not a tty
test_enosys: ioctl r=-1 errno=Function not implemented
'''
(From OE-Core rev: 2c5f70e7e73362502a7d8d0773fbd04520b26eef)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The 'cgroupv2' PACKAGECONFIG is dropped as systemd now only supports
cgroupv2.
The 'kmod' PACKAGECONFIG is now added 'libkmod' as a runtime dependency.
This is because systemd 256 uses dlopen to make use of libkmod, and there's
no way for our library detection codes to know this and add such dependency
automatically. So we add it explicitly.
mime is now inherited because ${datadir}/mime/packages/io.systemd.xml
is a mime type file.
For musl specific patches, most are refreshed. One old patch is dropped,
and four new ones are added.
Newly added:
build-path.c: avoid boot time segfault for musl
adjust header inclusion order to avoid redeclaration
undef stdin for references using stdin as a struct member
include signal.h to avoid the 'undeclared' error
Dropped:
src/basic/missing.h: check for missing strndupa
(drop reason: systemd does not use strndupa any more)
(From OE-Core rev: 89b75b46371d5e9172cb496b461824d8551a2af5)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix error in run.do_assemble_fitimage_initramfs.2779256:
line 238: [: =: unary operator expected
if FIT_CONF_DEFAULT_DTB is empty.
(From OE-Core rev: 19a6eea0951404403dcb5d0deeda8558b1337f82)
Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It looks like most of the assertions of this tests were by-passed
because of combinations of invalid regexes and inverted assert
statement. Try to fix.
Use the variables from the kernel recipe instead of potentially invalid
values from get_bb_vars with undefined recipe. Use one get_bb_vars
statement instead of many seaparate calls mainly to improve the duration
of this test case.
Make the test working for an its file with an absolute path to the
kernel binary. This will be needed with one of the following commits and
does not harm in general.
(From OE-Core rev: 741f0cfa4f0025a2823581cab09e0bf14382e54d)
Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Use more specific assert statements to make the code more readable and
the error messages more verbose.
Cleanup the asserts of the test case test_initramfs_bundle.
(From OE-Core rev: c7901004138d215612816072f20d367d638d2669)
Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Rather than writing hints into log files and verify the hints can be
found, the tests should verify that the artifacts in the deploy folder
are correctly signed. This is a much better test.
u-boot-tools provide a utility fit_check_sign which can verify the
signatures in fit images. Lets use it.
grepping in temp/run. or temp/log. files also does not work if the tasks
runs from sstate and the corresponding run file is not even generated.
(From OE-Core rev: 86e504b4f792eeadd67ea57dd71a62bcb4f16f02)
Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Eliminate code duplication with a new function.
Use get_bb_var instead of bitbake -e | grep.
(From OE-Core rev: 0c7b7c5c98217b5f139058533159f61d1989d46e)
Signed-off-by: Adrian Freihofer <adrian.freihofer@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When debug build is enabled(-Og is used), vulkan-samples do_compile
failed with error:
In function 'ZSTD_compressBlock_lazy_generic',
inlined from 'ZSTD_compressBlock_greedy' at TOPDIR/tmp-glibc/work/core2-32-wrs-linux/vulkan-samples/git/git/third_party/ktx/lib/basisu/zstd/zstd.c:21914:12:
TOPDIR/tmp-glibc/work/core2-32-wrs-linux/vulkan-samples/git/git/third_party/ktx/lib/basisu/zstd/zstd.c:21551:30: error: inlining failed in call to 'always_inline' 'ZSTD_HcFindBestMatch_selectMLS': function not considered for inlining
| FORCE_INLINE_TEMPLATE size_t ZSTD_HcFindBestMatch_selectMLS (
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
TOPDIR/tmp-glibc/work/core2-32-wrs-linux/vulkan-samples/git/git/third_party/ktx/lib/basisu/zstd/zstd.c:21736:32: note: called from here
| size_t const ml2 = searchMax(ms, ip, iend, &offsetFound);
Refer [1], always_inline is not suggested to use with indirect function
call, replace always_inline with inline to fix the issue
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107931
(From OE-Core rev: 246de52fe59de0612d1145357c5e904a51363c8c)
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The location of files changed during unpackdir transition. Adapt dirs
to fix custom icu data generation.
(From OE-Core rev: bb24c965a55d7e757c1e2a6e76bdb68d2bfb6ce3)
Signed-off-by: Stefan Müller-Klieser <s.mueller-klieser@phytec.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The current pciutils does not have this textrel problem any more.
(From OE-Core rev: 48fd3134097ce81c94c1ce67f218d83f1a5c2f6f)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It does not always make sense to collect artifacts and data from the
target on failure, e.g. if testing firmware or if the target is not
running an SSH server.
Allow this by setting TESTIMAGE_FAILED_QA_ARTIFACTS to an empty
value.
(From OE-Core rev: 7036f814baa6550e0b1ac2db37fcd51960ee9e48)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
