Repository navigation
Conversation
The executor strips ';' anyway, but with it the functions drop out of the task signature (OE-Core warns since nanbield). Checked with bitbake -e: the function sequence is identical before and after. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit bf7c578)
The executor strips ';' anyway, but with it the functions drop out of the task signature (OE-Core warns since nanbield). Checked with bitbake -e: the function sequence is identical before and after. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 27c2115)
The executor strips ';' anyway, but with it the functions drop out of the task signature (OE-Core warns since nanbield). Checked with bitbake -e: the function sequence is identical before and after. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 04752e2)
The executor strips ';' anyway, but with it the functions drop out of the task signature (OE-Core warns since nanbield). Checked with bitbake -e: the function sequence is identical before and after. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 8e5863c)
SRCPV is deprecated and expands to nothing; the fetcher adds the git revision to PKGV whenever PV contains "+git". Tested: bitbake -e easysplash gives the same PV, PKGV and PF before and after. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit ded797f)
The ConfigParsed handler wrote a tag into SRCPV:pn-<recipe> for the per-recipe check to read back, abusing a deprecated variable. The check now reads OSSYSTEMS_USE_ONSITE_ONLY_RECIPES itself. Tested: with easysplash listed, it is skipped for "0" and unset, parsed normally for "1". Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 958da17)
OE-Core made systemd the default INIT_MANAGER and dropped 'none', so oel-tiny gained 'systemd usrmerge' in DISTRO_FEATURES on a musl distro that runs tiny-init-system. Tested with bitbake -e: no systemd or usrmerge left, VIRTUAL-RUNTIME_* unchanged. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit cf14035)
e0d29f2 removed the OEL_DEFAULT_EXTRA_RRECOMMENDS hook from oel.inc, so oel-tiny's override became dead and, without systemd, oel.inc now recommends nss-myhostname, which is not buildable with musl. Tested with bitbake -e: DISTRO_EXTRA_RRECOMMENDS is empty for oel-tiny and oel. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 6937c11)
DISTRO_FEATURES_USBGADGET and DISTRO_FEATURES_WIFI are only set in
comments, so DISTRO_FEATURES carried them as literal unexpanded
'${...}' strings. Tested with bitbake -e for oel-tiny.
Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
(cherry picked from commit fa3a216)
OE-Core now expects distros to choose an init manager; oel has relied on its systemd default since wrynose. The busybox login_manager preference always lost to init-manager-systemd and every other init manager already defaults to busybox, so drop it. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 38041ff)
It was hooked up as DEFAULTTUNE_oel, pre-honister override syntax, so since 2021 it only set an unused variable and never changed any tune. Tested with bitbake -e busybox and core-image-minimal (qemuarm): only DEFAULTTUNE_oel and the function itself disappear. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 43a9f86)
The append always added a space, so a build without a build number shipped VERSION="26.11-snapshot-<date> ". Put the space inside the conditional string. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit bec260c)
1aa0efb moved cog-init with an extra cog/ level, so the recipe sat at recipes-browser/cog/cog/ and BBFILES_DYNAMIC (*/*/*.bb) never parsed it. Tested with a webkit stub layer: cog-init do_fetch, do_unpack and do_populate_lic succeed. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 99773ea)
BitBake warns about the missing whitespace on every parse with DISTRO=oel-tiny. Tested: bitbake -e output is identical apart from the two warnings, which are gone. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 64bb896)
…name
INITSCRIPT_NAME was "${BPM}", an undefined variable, so update-rc.d
registered a script literally named "${BPM}" and the installed
/etc/init.d/chromium-ozone-wayland never started. Tested: package_qa
with DISTRO=oel-tiny + sysvinit; the postinst now runs update-rc.d on
chromium-ozone-wayland.
Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
(cherry picked from commit ee591f1)
…ript meta-browser installs the browser only as /usr/bin/chromium; there is no /usr/bin/chromium-ozone-wayland, so the "[ -x $DAEMON ] || exit 0" guard made the sysvinit script a silent no-op. The systemd unit already runs /usr/bin/chromium. NAME stays, as it names the pidfile and the /etc/default file. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 69d09ed)
The script was installed as /etc/init.d/kiosk while INITSCRIPT_NAME, the systemd unit and the /etc/default file all use qt-kiosk-browser, so update-rc.d linked a script that does not exist. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 793842f)
…onfig NAME was "qt-kiosk-browser-init", so the script looked for /usr/bin/qt-kiosk-browser-init and /etc/qt-kiosk-browser-init.conf; neither exists, and "test -x || exit 0" made it a silent no-op. It also never read /etc/default/qt-kiosk-browser, which carries WAYLAND_DISPLAY, XDG_RUNTIME_DIR and QTWEBENGINE_CHROMIUM_FLAGS for the systemd unit. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit b4abb14)
OE-Core ships ppp 2.5.x, so the hard-coded plugin dir put the NM plugin where pppd never looks. meta-networking's PACKAGECONFIG[ppp] already takes the dir from pppd.pc. Tested: bitbake -e networkmanager with meta-networking master, default and with PACKAGECONFIG ppp. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 69ae464)
Upstream EasySplash has only a 1.0.0 tag, so the 2.0.0 recipes had no version to compare with. Tested with devtool check-upgrade-status: the three recipes went from UNKNOWN_BROKEN to new upstream commits. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit e8c462a)
The recipe needs ongoing work to stay buildable (its vendored Go module and license lists must be regenerated on every upgrade, and licenses did not resolve on non-x86 targets), and nothing in the layer or the platform uses it. Drop it instead of maintaining it. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 8bd5316)
Recipes whose only sources are files from this layer have no upstream
license file to checksum. OE-Core points LIC_FILES_CHKSUM at
${COMMON_LICENSE_DIR} for such recipes, but oelint-adv flags any
file:// path built from a variable other than WORKDIR, S or B, and the
rule is unchanged up to 9.11.3. Suppress it for the whole layer so
these recipes can use the common license text instead of shipping
their own copy. Recipes that fetch a source still checksum the license
file it ships.
Tested with contrib/oelint/run-oelint.sh (no findings).
Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
(cherry picked from commit d33f32c)
OE-Core now parses LICENSE as SPDX and warns on the legacy '|' operator. Tested with bitbake -p and easysplash do_populate_lic. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 10026f0)
LICENSE = "CLOSED" is deprecated. The recipe ships a config file from this layer, which is MIT licensed, so checksum the common MIT text instead of shipping a copy. Tested with do_populate_lic. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit df4d0a9)
LICENSE = "CLOSED" is deprecated. The recipe ships service, udev and config files from this layer, which is MIT licensed, so checksum the common MIT text instead of shipping a copy. Set S to UNPACKDIR, as the recipe has no source tree of its own. Tested with bitbake -p (oel skips the recipe: it needs wayland and pam). Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 906deab)
LICENSE = "CLOSED" is deprecated. The animation is built from the EasySplash repository, which is dual licensed with no separate terms for its data. The commercial flag stays for gstreamer1.0-libav. Tested with do_populate_lic. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit bf25684)
LICENSE = "CLOSED" is deprecated. The animation is built from the EasySplash repository, which is dual licensed with no separate terms for its data. The commercial flag stays for gstreamer1.0-libav. Tested with do_populate_lic. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 2cf15f4)
Upstream has no release tags, so the version check reported UNKNOWN_BROKEN. Tested with devtool check-upgrade-status: now MATCH. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 193de6c)
The tarball is named linuxconsoletools, not after PN, so the version check found nothing. Tested with devtool check-upgrade-status: 1.8.1, the latest release, now reports MATCH. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 2307819)
13.3 fixes the misspelt SysCtlN calls that broke writeback_limit, and drops the unsupported -s option from the help text. Tested with bitbake zram-init: buildhistory shows the same files and RDEPENDS, and only /usr/sbin/zram-init changes. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit b47a89c)
expand=True has long been the getVar() default, so passing it is noise. The getVar(x, False) calls in ossystems-distro-version stay as they disable expansion. Tested: bitbake -e with every touched class inherited gives the same expanded environment before and after. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit f72bb33)
Brings the systemd unit fixes (easysplash.enable=0 boot parameter, RemainAfterExit, no start in containers), the cargo dependency upgrade (gstreamer-rs 0.20; crates regenerated with update_crates) and the Makefile fix for the systemdsystemunitdir lookup (OSSystems/EasySplash#28). The recipe still passes systemdsystemunitdir explicitly so the unit directory does not come from pkg-config. Tested: bitbake easysplash and both animations on qemuarm; buildhistory shows only the expected deltas. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 7f36037)
The Makefile reads INITD, not INIT, so PACKAGECONFIG sysvinit never installed /etc/init.d/easysplash-start and update-rc.d had no script. Tested: bitbake easysplash on qemuarm, default oel (no change) and with PACKAGECONFIG sysvinit (init script installed, paths substituted). Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br> (cherry picked from commit 5cdfc2f)
lucianogdittgen
marked this pull request as ready for review
November 13, 2026 16:49
This was referenced Nov 16, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Backport to
wrynoseof the release-update PRs merged on master. Each commit is acherry-pick -x, in master merge order.Not backported, release-specific: #134 (LAYERSERIES_COMPAT and .oelint.cfg -> blacksail), #141 (DISTRO_VERSION 26.11-snapshot). #155's
.oelint.cfgconflicted only onrelease =, which stayswrynose. With this PR, wrynose differs from master only in those release settings and the root LICENSE file.Checked against OE-Core wrynose: ppp is 2.5.2 there too (so the 2.4.9 pin #146 drops is wrong on wrynose as well), init-manager-mdev-busybox.inc exists (#138), and Rust is 1.94.1, above the highest
rust-versionamong EasySplash 7829a12's crates (1.89, #159).Tested: oelint (
contrib/oelint/run-oelint.sh) passes. Not build-tested on wrynose (no wrynose build tree); the same commits are built on blacksail in #153.