summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* imx-pxp-g2d: Rename to imx-g2d-pxpg2d-renameOtavio Salvador34 hours3-5/+13
| | | | | | | | | | | | | | | | | | | | | | NXP is moving the G2D implementations to an imx-g2d-<backend> naming scheme, which groups them with imx-g2d-samples and matches the already-renamed nxp-imx/imx-g2d-pxp repository. Nothing changes for layers that go through virtual/libg2d. For the rest, the rename keeps the old name working: - PROVIDES the old name, so DEPENDS on imx-pxp-g2d still resolve. - RPROVIDES, RREPLACES and RCONFLICTS the old package, so package feeds and images built before the rename upgrade cleanly. - IMX_G2D_PXP_SRC defaults to the old IMX_PXP_G2D_SRC, so existing mirror overrides keep applying. A imx-pxp-g2d_%.bbappend in another layer must be renamed; BitBake reports it as dangling otherwise. PREFERRED_PROVIDER_virtual/libg2d set to the old name must be updated as well, since it is matched against the recipe name. Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
* imx-dpu-g2d: Rename to imx-g2d-dpuOtavio Salvador34 hours5-5/+13
| | | | | | | | | | | | | | | | | | | | | | NXP is moving the G2D implementations to an imx-g2d-<backend> naming scheme, which groups them with imx-g2d-samples and matches the already-renamed nxp-imx/imx-g2d-pxp repository. Nothing changes for layers that go through virtual/libg2d. For the rest, the rename keeps the old name working: - PROVIDES the old name, so DEPENDS on imx-dpu-g2d still resolve. - RPROVIDES, RREPLACES and RCONFLICTS the old package, so package feeds and images built before the rename upgrade cleanly. - IMX_G2D_DPU_SRC defaults to the old IMX_DPU_G2D_SRC, so existing mirror overrides keep applying. A imx-dpu-g2d_%.bbappend in another layer must be renamed; BitBake reports it as dangling otherwise. PREFERRED_PROVIDER_virtual/libg2d set to the old name must be updated as well, since it is matched against the recipe name. Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
* imx-gpu-g2d: Rename to imx-g2d-gpuOtavio Salvador34 hours3-5/+13
| | | | | | | | | | | | | | | | | | | | | | NXP is moving the G2D implementations to an imx-g2d-<backend> naming scheme, which groups them with imx-g2d-samples and matches the already-renamed nxp-imx/imx-g2d-pxp repository. Nothing changes for layers that go through virtual/libg2d. For the rest, the rename keeps the old name working: - PROVIDES the old name, so DEPENDS on imx-gpu-g2d still resolve. - RPROVIDES, RREPLACES and RCONFLICTS the old package, so package feeds and images built before the rename upgrade cleanly. - IMX_G2D_GPU_SRC defaults to the old IMX_GPU_G2D_SRC, so existing mirror overrides keep applying. A imx-gpu-g2d_%.bbappend in another layer must be renamed; BitBake reports it as dangling otherwise. PREFERRED_PROVIDER_virtual/libg2d set to the old name must be updated as well, since it is matched against the recipe name. Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
* Merge pull request #2750 from dolcini/claude/fix-fsl-tools-gpu-scopingOtavio Salvador36 hours1-0/+3
|\ | | | | packagegroup-fsl-tools-gpu: scope imx-gpu-viv-tools to Vivante GPUs
| * packagegroup-fsl-tools-gpu: scope imx-gpu-viv-tools to Vivante GPUsFrancesco Dolcini38 hours1-0/+3
|/ | | | | | | | | | | | | | | | | | | | | | | SOC_TOOLS_GPU:append:imxgpu unconditionally added imx-gpu-viv-tools for every imxgpu machine, but that tool package is only produced by imx-gpu-viv, whose COMPATIBLE_MACHINE does not cover non-Vivante imxgpu SoCs such as i.MX 95's Mali GPU. That left packagegroup-fsl-tools-gpu unbuildable there, breaking 'bitbake world'. Same issue class as PR #2748, just a different recipe it didn't touch. Move imx-gpu-viv-tools to its own SOC_TOOLS_GPU:append:imxviv, the same override imx-gpu-viv itself is gated on. imx-gpu-sdk stays under imxgpu, since it now builds on any GPU thanks to #2748's PACKAGECONFIG conversion. Tested with 'bitbake -n packagegroup-fsl-tools-gpu' for imx8mp-lpddr4-evk and imx95-19x19-verdin: i.MX 8M Plus keeps imx-gpu-viv-tools, i.MX 95 now resolves instead of failing with "Nothing RPROVIDES 'imx-gpu-viv-tools'". Assisted-by: LLM Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
* Merge pull request #2748 from Freescale/imx-gpu-opencl-openvg-scopingOtavio Salvador3 days4-85/+101
|\ | | | | Fix OpenCL/OpenVG provider scoping for non-OpenCL Vivante and Mali GPUs
| * imx-gpu-sdk: Convert the GPU features to PACKAGECONFIGimx-gpu-opencl-openvg-scopingOtavio Salvador3 days1-58/+69
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The recipe assumed a Vivante GPU on every imxgpu machine: OpenVG and HW_GPU_VIVANTE were always passed to FslBuild.py and virtual/libopenvg was a dependency for all 2D GPUs. OpenVG is only implemented by the Vivante driver, so on i.MX 95, with its Mali GPU, nothing provides virtual/libopenvg and even with the dependency dropped the build would still request the OpenVG samples and the Vivante-specific code. Follow the layout used by NXP's meta-imx for this recipe and describe each FslBuild.py feature as a PACKAGECONFIG option carrying its own dependencies, so the features passed to the build and the recipe dependencies can no longer diverge: - openvg and vivante are only enabled on imxviv, mali on imxmali; - i.MX 95 gets the OpenGL ES 3.x, OpenCV and Vulkan features and the GL extensions list used by meta-imx, plus OpenCL when the opencl distro feature is enabled; - Vulkan is enabled only with the vulkan distro feature, as vulkan-loader and rapidvulkan require it; - the Vivante machines keep the same feature set as before. libxdg-shell is now a runtime dependency for every Wayland build, not only on i.MX 8, matching its build dependency. Tested with 'bitbake -e' and 'bitbake -n imx-gpu-sdk' for imx6qdlsabresd, imx7ulpevk, imx8mm-lpddr4-evk, imx8mp-lpddr4-evk, imx8mq-evk and imx95-19x19-verdin, with and without the opencl distro feature, comparing FEATURES against the previous recipe for the Vivante machines. Reported-by: Francesco Dolcini <francesco.dolcini@toradex.com> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
| * opencv: Follow IMXGPU_OPENCL_PROVIDER for OpenCL supportOtavio Salvador3 days1-2/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | OpenCL was enabled for every i.MX GPU but i.MX 8M Mini. On i.MX 7ULP the Vivante driver has no OpenCL either, and on i.MX 95 the Mali GPU relies on the standard ICD loader, which requires the opencl distro feature; without it virtual/libopencl1 has no provider and opencv cannot be built. Enable OpenCL on Vivante GPUs only when IMXGPU_OPENCL_PROVIDER is set, which covers the i.MX 8M Mini case without a SoC-specific override, and on other GPUs only when the opencl distro feature is enabled. Tested with 'bitbake -n opencv' for imx7ulpevk, imx8mm-lpddr4-evk, imx8mp-lpddr4-evk, imx8mq-evk and imx95-19x19-verdin, with and without the opencl distro feature. Reported-by: Francesco Dolcini <francesco.dolcini@toradex.com> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
| * imx-base.inc: Add IMXGPU_OPENCL_PROVIDER to select the OpenCL providerOtavio Salvador3 days2-25/+27
|/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The OpenCL preferred providers were set to imx-gpu-viv for every Vivante GPU machine, while imx-gpu-viv itself cleared its OpenCL PROVIDES and packages for i.MX 7ULP and i.MX 8M Mini, whose GPU cores lack OpenCL support. With PREFERRED_PROVIDER_virtual/libopencl1 naming a recipe that does not provide it, base.bbclass skips the remaining providers (opencl-icd-loader, ocl-icd), leaving virtual/libopencl1 unbuildable and breaking 'bitbake world' on those machines. The runtime side had the same mismatch, as PREFERRED_RPROVIDER of virtual-opencl-icd pointed to libopencl-imx, which is not produced there. Add IMXGPU_OPENCL_PROVIDER, set to imx-gpu-viv for Vivante GPUs and cleared for i.MX 7ULP and i.MX 8M Mini, as the single place deciding it: - the OpenCL PREFERRED_PROVIDER entries use it, falling back to the standard recipes (and opencl-icd-loader, as meta-oe does, for virtual/libopencl1) when it is empty; - PREFERRED_RPROVIDER of virtual-opencl-icd falls back to the default libopencl-mesa; - imx-gpu-viv derives its OpenCL PROVIDES and packages from it instead of keeping its own SoC list, and the libopencl-imx RPROVIDES overrides for packages that no longer exist are dropped. Machines can now also opt out of the Vivante OpenCL stack by clearing IMXGPU_OPENCL_PROVIDER. Tested with 'bitbake -e' and 'bitbake -n' for imx6qdlsabresd, imx7ulpevk, imx8mm-lpddr4-evk, imx8mp-lpddr4-evk, imx8mq-evk and imx95-19x19-verdin, with and without the opencl distro feature: i.MX 6, 8M Plus and 8M Quad keep imx-gpu-viv as provider, while i.MX 7ULP and 8M Mini now resolve clinfo, piglit and opencl-icd-loader through the standard loader. Reported-by: Francesco Dolcini <francesco.dolcini@toradex.com> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
* Merge pull request #2745 from Freescale/socarch-filter-qtdeclarativeOtavio Salvador3 days1-0/+3
|\ | | | | imx-base.inc: Add Qt modules to the SoC arch filters
| * imx-base.inc: Add Qt modules to the i.MX GPU SoC arch filtersocarch-filter-qtdeclarativeOtavio Salvador3 days1-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | On the NXP BSP, qtbase moves to MACHINE_SOCARCH through its GL dependencies, but qtbase itself is not in the imxgpu filter. Every Qt module built on it therefore keeps the tune architecture: qtdeclarative, qtsvg, qtxmlpatterns, qtgraphicaleffects, qtquickcontrols and qtquickcontrols2 are all built against the SoC-specific qtbase and still share their package with every other machine of the same tune. fsl-dynamic-packagearch only matches a recipe's own PROVIDES and direct DEPENDS, so both levels have to be listed. qtbase covers the modules that depend on it directly, and qtdeclarative covers the QML modules that depend only on it, the same pair the mainline BSP filter uses. Tested by evaluating PACKAGE_ARCH for imx8mm-lpddr4-evk and imx6qdlsabresd with IMX_DEFAULT_BSP set to nxp. Before this change only qtbase is cortexa53-crypto-mx8mm and cortexa9hf-neon-mx6qdl, and the other six modules keep cortexa53-crypto and cortexa9hf-neon; with it, all seven follow qtbase. Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
| * imx-base.inc: Add qtdeclarative to the mainline BSP SoC arch filterOtavio Salvador3 days1-0/+1
|/ | | | | | | | | | | | | | | | | | | | | | | | | | | | fsl-dynamic-packagearch only matches a recipe's own PROVIDES and direct DEPENDS against MACHINE_SOCARCH_FILTER. qtbase is in the filter for the mainline BSP, so qtdeclarative, qtsvg and qtxmlpatterns, which depend on it directly, move to MACHINE_SOCARCH. qtgraphicaleffects, qtquickcontrols and qtquickcontrols2 depend only on qtdeclarative and keep the tune architecture, although they are built against the SoC-specific qtbase. A build directory shared with a non-i.MX machine of the same tune then produces two different builds of those packages under one name and version, and whichever ran last overwrites the other. buildhistory reports it as version-going-backwards, since the PR service numbers the two builds differently. Adding qtdeclarative moves every QML module that depends on it to MACHINE_SOCARCH. Tested on a wrynose tree, building an image for an i.MX7D machine (imx-mainline-bsp) and then for an Allwinner A40i machine with the same cortexa7hf-neon tune, followed by the i.MX7D image again, in one build directory. That order failed before this change; it now succeeds, and the i.MX7D image ships qtdeclarative, qtgraphicaleffects, qtquickcontrols and qtquickcontrols2 as cortexa7hf-neon-imx. Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
* Merge pull request #2743 from YoeDistro/yoe/mutOtavio Salvador5 days3-1/+51
|\ | | | | Support blacksail and fix build with clang
| * imx-dpu-g2d: Fix strtol() overflow check in getSystemEnvInt()Khem Raj5 days2-0/+50
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | getSystemEnvInt() stores the strtol() result in an int and then compares it against LONG_MAX/LONG_MIN. On LP64 the value is truncated first, so the ERANGE check can never trigger, and clang fails the build: os/linux/g2d_utils.c:33:31: error: result of comparison of constant 9223372036854775807 with expression of type 'int' is always false [-Werror,-Wtautological-constant-out-of-range-compare] Add a patch that keeps the value in a long, rejects values outside the int range and narrows explicitly on return. Tested with bitbake imx-dpu-g2d on imx8qm-var-som using clang. Signed-off-by: Khem Raj <khem.raj@oss.qualcomm.com>
| * layer: Add blacksail YP 6.1 to compatible releasesKhem Raj5 days1-1/+1
|/ | | | Signed-off-by: Khem Raj <raj.khem@gmail.com>
* Merge pull request #2741 from dolcini/dev/fix-preferred-version-libcamera-imxOtavio Salvador6 days1-4/+4
|\ | | | | imx-base: fix stale PREFERRED_VERSION_LIBCAMERA_IMX after 0.7.1 bump
| * imx-base.inc: fix stale PREFERRED_VERSION_LIBCAMERA_IMX after 0.7.1 bumpFrancesco Dolcini7 days1-4/+4
| | | | | | | | | | | | | | | | | | | | The 0.6.0 -> 0.7.1 bump renamed libcamera_0.6.0.imx.bb to libcamera_0.7.1.imx.bb but left PREFERRED_VERSION_LIBCAMERA_IMX pointing at the now-nonexistent "0.6.0.imx" for mx95/mx8mm/mx8ulp/ mx8mq-nxp-bsp. Bump it to "0.7.1.imx" to match. Fixes: f7e2763577d1 ("libcamera: Bump 0.6.0 -> 0.7.1") Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
* | Merge pull request #2740 from nxp-upstream/wifi-firmware-updateOtavio Salvador6 days1-2/+13
|\ \ | | | | | | firmware-nxp-wifi: Add nxpaw693 sdio module
| * | firmware-nxp-wifi: Add nxpaw693 sdio moduleZelan Zou8 days1-2/+13
|/ / | | | | | | | | | | upgrade to lf-6.18.20-2.0.0 release Signed-off-by: Zelan Zou <zelan.zou@nxp.com>
* | Merge pull request #2739 from lucianogdittgen/ci-oelint-workflowRodrigo Medeiros8 days3-0/+82
|\ \ | | | | | | oelint: Add CI workflows for recipe linting
| * | oelint-homepage: Report unreachable HOMEPAGE values weeklyLuciano Dittgen13 days1-0/+40
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | oelint.vars.homepageping is excluded from the blocking gate because a shared runner egress IP gets rate-limited and reports live URLs as dead; four full runs here disagreed with each other on an unchanged tree. Link rot is still worth knowing about, so run the rule weekly instead, reporting into the run summary and never failing a pull request. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
| * | oelint: Lint recipes with oelint-adv on pull requestsLuciano Dittgen13 days2-0/+42
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Rule-by-rule remediation took this layer from 51 findings to zero, but nothing stops a new recipe reintroducing them. Findings become GitHub annotations on the diff. oelint.vars.homepageping is excluded: it pings 105 github.com HOMEPAGEs per run until a shared runner IP is rate-limited into false positives. All three oelint packages are pinned, since oelint-adv's own ranges float. Verified clean on master plus #2730. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | Auto-update LICENSE file with current recipe licensesgithub-actions[bot]10 days1-1/+0
| | |
* | | Merge pull request #2730 from lucianogdittgen/testfloat-use-meta-oe-3eRodrigo Medeiros10 days5-3190/+0
|\ \ \ | |/ / |/| | testfloat: Drop the 2a pin and use meta-oe's TestFloat 3e
| * | testfloat: Drop the 2a pin and use meta-oe's TestFloat 3eLuciano Dittgen13 days5-3190/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The bundled 2a recipe cannot build on any current toolchain: its SPE patch calls __builtin_spe_mfspefscr, removed in GCC 9, and includes bits/nan.h, removed in glibc 2.26. qoriq-base.inc is included by the arm, arm64 and ppc configs, so the pin also blocked meta-oe's portable testfloat_3e on machines 2a could never build for anyway. Tested with bitbake testfloat on p2020rdb: 3e builds and ships five binaries. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | Merge pull request #2734 from lucianogdittgen/oelint-vars-descriptiontoobriefOtavio Salvador13 days1-1/+4
|\ \ \ | | | | | | | | gst-examples: Expand DESCRIPTION beyond SUMMARY
| * | | gst-examples: Expand DESCRIPTION beyond SUMMARYLuciano Dittgen13 days1-1/+4
| |/ / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | DESCRIPTION restated the recipe name in fewer words than SUMMARY, so it told a reader less than the line above it; name the two example programs instead. The file is a downgraded copy of the oe-core recipe but not a verbatim fork -- master already carries local SECTION, DEPENDS order and indent changes to it -- so it is maintained metadata. Built gst-examples for imx8mm-lpddr4-evk: no buildhistory delta, new text in pkgdata. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | Merge pull request #2721 from lucianogdittgen/oelint-task-pythonprefixOtavio Salvador13 days1-0/+3
|\ \ \ | | | | | | | | jailhouse-imx: Suppress pythonprefix on the shell compile prepend
| * | | jailhouse-imx: Suppress pythonprefix on the shell compile prependLuciano Dittgen2026-09-171-0/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | oelint.task.pythonprefix parses a task body with ast and reports it when it happens to be valid Python. This prepend matches only because a lone oe_runmake is also a Python expression statement; the rule's own line-count guard is meant to exempt that, but counts the body's comment as code. The task must stay shell. bitbake -e shows setuptools3 is inherited after module and shadows module_do_compile, so this prepend builds the module. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | | Auto-update LICENSE file with current recipe licensesgithub-actions[bot]13 days1-1/+1
| | | |
* | | | Merge pull request #2718 from lucianogdittgen/oelint-var-inheritdevtool-nativeOtavio Salvador13 days1-4/+2
|\ \ \ \ | | | | | | | | | | imx-mkimage: Rename to imx-mkimage-native and defer the native inherit
| * | | | imx-mkimage: Defer the native inheritLuciano Dittgen2026-09-171-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | devtool inherits extra classes after the recipe, which can push native out of the last position bitbake expects. Deferring keeps it last regardless of inherit order. class-native, PACKAGE_ARCH and the empty STAGING_DIR_TARGET are unchanged. Tested with bitbake imx-mkimage-native on imx8mp-lpddr4-evk: empty buildhistory delta. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
| * | | | imx-mkimage: Rename the recipe to imx-mkimage-nativeLuciano Dittgen2026-09-171-3/+0
| |/ / / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The recipe is native-only, but PN did not end in -native, so native.bbclass's virtclass handler returned before applying its DEPENDS remapping. BPN prunes the suffix via SPECIAL_PKGSUFFIX, so a -native filename cannot double it, contrary to the note bb3945d1 added and this drops. No consumer referenced the old PN. Tested with bitbake imx-mkimage-native on imx8mp-lpddr4-evk: sysroot contents unchanged. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | | Merge pull request #2727 from lucianogdittgen/oelint-spaces-lineendOtavio Salvador13 days1-0/+5
|\ \ \ \ | | | | | | | | | | u-boot-fslc: Document why the conditional inherit cannot be deferred
| * | | | u-boot-fslc: Document why the conditional inherit cannot be deferredLuciano Dittgen2026-09-171-0/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The line trips two rules. inherit_defer would run imx-boot-container at end of parse, after uuu_bootloader_tag, whose do_deploy:append tags flash.bin before the symlink exists -- do_deploy fails that way on imx8mq-evk. The trailing space is only in the linter's empty-branch expansion; no line in the file ends in whitespace. Deferring both inherits just moves the finding to the bare inherit_defer below. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | | | Merge pull request #2729 from lucianogdittgen/oelint-vars-inconspaces-g2dOtavio Salvador13 days2-2/+2
|\ \ \ \ \ | |_|_|/ / |/| | | | imx-g2d: Set S directly instead of appending to it
| * | | | imx-gpu-g2d: Set S directly instead of appending to itLuciano Dittgen13 days1-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | S is a path, so the :append joined without a separator and tripped oelint.vars.inconspaces, whose leading space would have broken it. A direct assignment drops the construct and matches how every other recipe in the layer points at a source subdirectory. Tested with bitbake imx-gpu-g2d on imx8mm-lpddr4-evk: S expands byte-identically and the buildhistory delta is empty. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
| * | | | imx-dpu-g2d: Set S directly instead of appending to itLuciano Dittgen13 days1-1/+1
|/ / / / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | S is a path, so the :append joined without a separator and tripped oelint.vars.inconspaces, whose leading space would have broken it. A direct assignment drops the construct and matches how every other recipe in the layer points at a source subdirectory. Tested with bitbake imx-dpu-g2d on imx95-19x19-verdin: S expands byte-identically and the buildhistory delta is empty. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | | Auto-update LICENSE file with current recipe licensesgithub-actions[bot]14 days1-3/+3
| | | |
* | | | Merge pull request #2728 from ↵Otavio Salvador14 days9-150/+147
|\ \ \ \ | |/ / / |/| | | | | | | | | | | eichenberger/sync-g2d-with-meta-imx-lf-6.18.20_2.0.0 Sync g2d with meta imx lf-6.18.20_2.0.0
| * | | imx-pxp-g2d: Update to the BSD-3-Clause licensed revisionStefan Eichenberger2026-09-171-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Bump SRCREV to the head of the imx_2.5 branch. Upstream relicensed the library from NXP's proprietary license to BSD-3-Clause there; apart from the license headers and an updated SBOM the sources are unchanged. Signed-off-by: Stefan Eichenberger <stefan.eichenberger@toradex.com>
| * | | imx-gpu-g2d: Build from source instead of prebuilt binaryStefan Eichenberger2026-09-172-45/+51
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Replace the prebuilt binary recipe with one that builds imx-gpu-g2d from the upstream nxp-imx/imx-gpu-g2d source, tracking the imx_2.5 branch. The new version of imx-gpu-g2d is BSD-3-Clause licensed while the old version was using a proprietary license. As a result the recipe no longer needs the NXP EULA to be accepted. Signed-off-by: Stefan Eichenberger <stefan.eichenberger@toradex.com>
| * | | imx-dpu-g2d: Build from source instead of prebuilt binaryStefan Eichenberger2026-09-173-53/+86
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Replace the prebuilt binary recipe with one that builds imx-dpu-g2d from the upstream nxp-imx/imx-dpu-g2d source, tracking the imx_2.5 branch. Carry a local patch to drop an unused variable. The new version of imx-dpu-g2d is BSD-3-Clause licensed while the old version was using a proprietary license. Signed-off-by: Stefan Eichenberger <stefan.eichenberger@toradex.com>
| * | | linux-imx: Update to 6.18.20 (lf-6.18.y)Stefan Eichenberger2026-09-173-49/+7
|/ / / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Bump SRCREV to the latest lf-6.18.y tip and LINUX_VERSION to 6.18.20, bumping LOCALVERSION for the ABI change. Drop the ipu uapi integer types patch. Upstream now includes <stdint.h> in include/uapi/linux/ipu.h and no longer defines uint32_t, uint16_t and uint8_t as macros, which is what the patch was carried for. The header still defines u32, u8 and __u32 as macros, so including <linux/types.h> after it remains unwise, but the standard integer type names are no longer affected. Also install the imx_drm.h uapi header via linux-imx-headers. It is needed by the imx-dpu-g2d recipe, which builds from source and includes <imx/drm/imx_drm.h>. Signed-off-by: Stefan Eichenberger <stefan.eichenberger@toradex.com>
* | | Merge pull request #2722 from lucianogdittgen/oelint-file-inlinesuppress_naOtavio Salvador2026-09-171-0/+9
|\ \ \ | | | | | | | | oelint: Suppress inlinesuppress_na, which misreports live suppressions
| * | | oelint: Suppress inlinesuppress_na, which misreports live suppressionsLuciano Dittgen2026-09-171-0/+9
| |/ / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | oelint.file.inlinesuppress_na reported all six of this layer's inline suppressions as unneeded. Each was checked by deleting it and re-scanning: every one brought back the finding it was said not to suppress, and optee-os-fslc.inc alone brought back six oelint.var.override errors. Acting on the rule makes the layer worse, so suppress it layer-wide and re-evaluate on an oelint-adv newer than 9.11.2. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | Merge pull request #2720 from lucianogdittgen/oelint-vars-pythonrdependsOtavio Salvador2026-09-171-1/+1
|\ \ \ | | | | | | | | ovs-dpdk: Drop the unused python3 runtime dependency
| * | | ovs-dpdk: Drop the unused python3 runtime dependencyLuciano Dittgen2026-09-171-1/+1
| |/ / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | do_install:append installs seven compiled binaries and a JSON schema, and the recipe inherits no autotools, so base_do_install stays empty: the package ships no Python at all and python3 only pulled the interpreter and its stdlib onto the target. Tested with bitbake ovs-dpdk on ls1046ardb: FILELIST and PKGSIZE unchanged, ovs-dpdk-dev loses python3-dev. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
* | | Merge pull request #2719 from lucianogdittgen/oelint-vars-homepagepingOtavio Salvador2026-09-171-1/+1
|\ \ \ | | | | | | | | glslang: Point HOMEPAGE at the upstream repository
| * | | glslang: Point HOMEPAGE at the upstream repositoryLuciano Dittgen2026-09-171-1/+1
| |/ / | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The Khronos Reference-Compiler page has been retired and answers 404, so the recipe advertised a dead link. The GitHub repository is the project's canonical home, is already this recipe's SRC_URI, and matches the sibling spirv-tools downgrade. oe-core carries the same dead URL and wants the same fix. Tested with bitbake glslang on imx8mp-lpddr4-evk: empty buildhistory delta. Signed-off-by: Luciano Dittgen <luciano.dittgen@ossystems.com.br>
generated by cgit v1.2.3-54-g00ecf (git 2.39.0) at 2026-10-01 23:48:54 +0000