summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Merge pull request #2925 from mdrodrigo/imx952-sm-basenameHEADmasterOtavio Salvador7 hours1-0/+2
|\ | | | | imx952-evk.inc: Set the System Manager firmware base name
| * imx952-evk.inc: Set the System Manager firmware base nameRodrigo M. Duarte10 hours1-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The i.MX 952 machines never set SYSTEM_MANAGER_FIRMWARE_BASENAME, so imx-system-manager tried to deploy build/mx952evk/.bin and failed: install: cannot stat '.../build/mx952evk/.bin': No such file or directory Set it to m33_image, as the NXP BSP does for i.MX 952 and as imx943-evk.inc and imx95-19x19-verdin.conf already do. Based on LF6.18.37_2.1.0 NXP BSP release. Tested by building imx-system-manager and imx-boot with the NXP BSP for imx952-15x15-lpddr5-frdm and imx952-19x19-lpddr5-evk: both deploy m33_image-mx952evk.bin and build the flash_a55 boot container. The default flash_all target also needs the M7 image in the boot container, which imx-boot does not stage yet. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* | Merge pull request #2924 from mdrodrigo/imx95-b1Otavio Salvador7 hours1-1/+1
|\ \ | |/ |/| imx-base: Install the i.MX 95 B1 ELE runtime firmware
| * imx-base: Install the i.MX 95 B1 ELE runtime firmwareRodrigo M. Duarte14 hours1-1/+1
|/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The NXP 2.1.0 BSP adds support for i.MX 95 B1 silicon. The ELE firmware in firmware-ele-imx 2.0.7 already ships the B1 boot container (mx95b1-ahab-container.img) and the B1 runtime container (mx95b1runtime-ahab-container.img). The boot container name already follows IMX_SOC_REV, and imx-mkimage and imx-oei treat REV=B1 like B0. The runtime containers, however, are installed from a fixed list, which lacks B1. Add the B1 runtime container to that list so that a rootfs works on A0, B0 and B1 silicon. Unlike the NXP BSP, the A0 runtime container is kept, as boards may still set IMX_SOC_REV to A0. A board with B1 silicon sets IMX_SOC_REV = "B1" in its machine file. NXP's imx95-b1-* machines are not added, as they require the i.MX 95 EVK and FRDM machines, which meta-freescale does not have. With the mainline BSP, u-boot-fslc builds the boot container itself, and its imx95-u-boot.dtsi always uses the B0 ELE container, so B1 is only supported with the NXP BSP for now. Tested by building firmware-ele-imx and imx-boot for imx95-19x19-verdin with the NXP BSP, with IMX_SOC_REV set to B1 and with the default B0: each imx-boot contains the ELE container for its revision, and firmware-ele-imx installs the A0, B0 and B1 runtime containers. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* Merge pull request #2923 from metamacro/fix/imx-gst1.0-plugin-libdrmOtavio Salvador15 hours1-1/+1
|\ | | | | imx-gst1.0-plugin: Always depend on libdrm
| * imx-gst1.0-plugin: Always depend on libdrmAndrej Gardijan15 hours1-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | plugins/compositor, plugins/videoconvert and tools/grecorder include libdrm/drm_fourcc.h on every platform. The recipe only depended on libdrm for imxfbdev:imxgpu. A distro without wayland and opengl (e.g. an i.MX 91 audio device) fails in do_compile: gstimxcompositor.c:125:10: fatal error: libdrm/drm_fourcc.h: No such file or directory Signed-off-by: Andrej Gardijan <andrej.gardijan@zenitel.com>
* | Merge pull request #2922 from mdrodrigo/opencv-fastcv-depsOtavio Salvador15 hours1-3/+3
|\ \ | | | | | | opencv: Correct FastCV dependencies
| * | opencv: Correct FastCV dependenciesRodrigo M. Duarte32 hours1-3/+3
| |/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Apply meta-openembedded commit bd832a8962 ("opencv: correct FastCV dependencies") to the meta-openembedded copy, as NXP release LF6.18.37_2.1.0 does in meta-imx commit be82d45d76 ("opencv: Align with OE-Core recipe"). The FastCV HAL dlopen()s libfastcvopt.so.1 and works without it, so qcom-fastcv-binaries becomes an RRECOMMENDS of libopencv-core, libopencv-fastcv and libopencv-imgproc instead of an RDEPENDS. The dependency is also dropped from PACKAGECONFIG[fastcv], since populate_packages already adds it. This lets the fastcv PACKAGECONFIG build without meta-qcom, the only layer providing qcom-fastcv-binaries. bd832a8962 is the only meta-openembedded opencv change after the previous copy, so the Upstream hash moves to it. fastcv is not in the default PACKAGECONFIG, so default builds are unchanged. Tested by building opencv for imx8mq-evk with fastcv added to PACKAGECONFIG and without meta-qcom: it configures with WITH_FASTCV=ON, builds libopencv-fastcv, and libopencv-core, libopencv-fastcv and libopencv-imgproc list qcom-fastcv-binaries only in RRECOMMENDS. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* | Merge pull request #2919 from dolcini/dev/imx8-scu-no-sdma-firmwareOtavio Salvador15 hours1-3/+3
|\ \ | |/ |/| imx-base: Drop the SDMA firmware on i.MX 8QuadMax/8QuadXPlus/8DualX
| * imx-base: Drop the SDMA firmware on i.MX 8QuadMax/8QuadXPlus/8DualXFrancesco Dolcini46 hours1-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | The SDMA firmware was added for all the i.MX 8, but i.MX 8QuadMax, 8QuadXPlus and 8DualX have no SDMA controller. Their reference manuals (i.MX 8QuadMax and i.MX 8DualX/8DualXPlus/8QuadXPlus Applications Processor Reference Manuals) list only eDMA controllers, and in the Linux device tree (imx8qm.dtsi, imx8qxp.dtsi, imx8dx.dtsi and the imx8-ss-*.dtsi subsystems they include) the DMA controllers are eDMA, with no SDMA node. The firmware has no user. Fixes: 88f6c00d6814 ("imx-base.inc: Add linux-firmware-imx-sdma-imx7d for all i.MX 8") Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
* | Merge pull request #2920 from mdrodrigo/linux-imx-inmate-dtbOtavio Salvador38 hours4-16/+24
|\ \ | |/ |/| Linux imx inmate dtb
| * conf/machine: Use the renamed Jailhouse inmate device treesRodrigo M. Duarte39 hours3-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | NXP's 2.1.0 kernel replaces the per-board Jailhouse inmate device trees with one per SoC: imx8mp-evk-inmate.dtb -> imx8mp-jailhouse-inmate.dtb imx93-11x11-evk-inmate.dtb -> imx93-jailhouse-inmate.dtb imx95-19x19-verdin-inmate.dtb -> imx95-jailhouse-inmate.dtb The last one comes from linux-imx commit 45637d373731 ("LF-17737-2 arm64: dts: imx95: Unify i.MX95 inmate dts"). Both NXP BSP kernels, linux-imx 6.18.37 and linux-fslc-imx 6.18.56, now build only the new names, so the kernel build fails on the old ones. Update imx8mp-lpddr4-evk, imx93-11x11-lpddr4x-evk and imx95-19x19-verdin, as meta-imx does for the first two. The entries are NXP BSP only, so the mainline BSP is not affected. Tested by building the kernel for the three machines with both linux-imx and linux-fslc-imx; each deploys the renamed inmate DTB along with every other device tree in KERNEL_DEVICETREE. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
| * linux-fslc-imx: Bump kernel to LF6.18.37_2.1.0 + stable up to v6.18.56Rodrigo M. Duarte39 hours1-13/+21
|/ | | | | | | | | | | | | | | | | | | | | Move to the 6.18-2.1.x-imx branch of Freescale/linux-fslc, which merges NXP's lf-6.18.37-2.1.0 (the current head of lf-6.18.y) and stable up to v6.18.56. Update KBRANCH, SRCREV and LINUX_VERSION accordingly, and refresh the tracking header: the stable and NXP tags, and the commits carried on top of both, listed with git log --oneline --no-merges v6.18.56.. ^linux-6.18.y ^NXP/lf-6.18.y This also brings the device tree renames from NXP's 2.1.0 release, which the following commit adapts the machine configurations to. Tested on imx8mq-evk with the NXP BSP: core-image-weston builds and the out-of-tree galcore, mlan and moal modules match the kernel release 6.18.56-lf-6.18.y-g4ce72be4ed55. The kernel and all listed device trees also build for imx8mp-lpddr4-evk, imx93-11x11-lpddr4x-evk and imx95-19x19-verdin. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* Merge pull request #2918 from wardru/fix/imx93-boot-container-mainlineOtavio Salvador2 days1-2/+2
|\ | | | | imx-base: Restrict the i.MX93 U-Boot boot container to mainline
| * imx-base: Restrict the i.MX93 U-Boot boot container to mainlineAntoine Gouby2 days1-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Commit ece8c0bf0705 meant to enable the U-Boot boot container for mainline i.MX93 only, but used the mx93-generic-bsp override, which matches all i.MX93 machines unless IMX_DEFAULT_BOOTLOADER is u-boot-imx. If a vendor uses a U-Boot other than u-boot-imx that does not build the boot container , do_deploy fails: install: cannot stat '.../build/<config>-<type>/flash.bin': No such file or directory Use the mx93-mainline-bsp override, as done for i.MX95, to limit the U-Boot boot container to mainline builds. Fixes: ece8c0bf0705 ("imx-base: Use the U-Boot boot container for mainline i.MX 93") Signed-off-by: Antoine Gouby <antoine.gouby@toradex.com>
* | Merge pull request #2917 from wardru/fix/pipewireOtavio Salvador2 days2-55/+0
|\ \ | |/ |/| pipewire: Drop per-plane stride and offset patch for iMX95
| * pipewire: Drop per-plane stride and offset patch for i.MX95Antoine Gouby3 days2-55/+0
|/ | | | | | | | | The patch was merged upstream as commit 95e89f786aad ("pipewiresrc: update per-plane stride and offset according to chunk info") and backported to the 1.6 branch as commit eb12bca634eb, which is part of the 1.6.10 release onwards. Drop it as it is no longer needed. Signed-off-by: Antoine Gouby <antoine.gouby@toradex.com>
* Merge pull request #2915 from dsd/masterOtavio Salvador6 days5-6/+8
|\ | | | | Enable mainline bootloader, kernel and Open NPU stack on FRDM-IMX93
| * mesa: Enable the Ethos-U driver and Teflon delegate on i.MX93Daniel Drake6 days1-0/+1
| | | | | | | | | | | | | | | | | | | | | | With the ethosu kernel driver enabled in linux-fslc, the i.MX93 Ethos-U65 NPU can now be used through mainline kernel + mesa, without needing any proprietary vendor NPU stack components. Enable the relevant mesa driver and delegate to enable NPU usage via TensorFlow Lite. Signed-off-by: Daniel Drake <ddrake@baylibre.com>
| * linux-fslc: Update to enable the Ethos-U NPU driverDaniel Drake6 days1-1/+1
| | | | | | | | | | | | | | | | | | | | Update to the latest 7.2.x+fslc, enabling the Arm Ethos-U NPU driver on i.MX93. Relevant changes: - 672fe6502b73 arm64: defconfig: Enable Arm Ethos-U NPU driver Signed-off-by: Daniel Drake <ddrake@baylibre.com>
| * imx93-11x11-lpddr4x-frdm: Allow mainline kernel and bootloaderDaniel Drake6 days1-3/+1
| | | | | | | | | | | | | | | | | | | | | | | | This machine config was forcing use of the vendor kernel/bootloader via PREFERRED_PROVIDER variables. However, the recent fslc upgrades to Linux 7.2 and u-boot 2026.10 include decent mainline support for this device. Drop the overrides and select the appropriate mainline u-boot build config so that the mainline BSP can be used on this device. Signed-off-by: Daniel Drake <ddrake@baylibre.com>
| * machine/imx93-evk: Allow mainline BSP to be selectedDaniel Drake6 days1-2/+4
| | | | | | | | | | | | | | | | | | | | | | | | The i.MX 93 machine include sets IMX_DEFAULT_BSP = "nxp" in a way that overrides the "mainline" default in imx-base.inc, and also overrides any user preference selected by e.g. enabling the bsp/fslc fragment. This is despite some level of mainline support being available. Instead of the hard override, make "nxp" a soft default, which allows mainline to be selected if desired. Signed-off-by: Daniel Drake <ddrake@baylibre.com>
| * imx-base: Use the U-Boot boot container for mainline i.MX 93Daniel Drake6 days1-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Mainline U-Boot for i.MX 93 assembles flash.bin itself with binman, and fails to build when the DDR, ELE and TF-A binaries are not in its build directory: Image 'u-boot-spl-ddr' is missing external blobs and is non-functional imx-boot-container already supports i.MX 93: it stages those binaries and deploys flash.bin as imx-boot. Enable it for mainline builds as is done for i.MX 8M. Signed-off-by: Daniel Drake <ddrake@baylibre.com>
* | Merge pull request #2916 from mdrodrigo/jailhouse-imx-2.1.0Otavio Salvador6 days1-3/+8
|\ \ | |/ |/| Jailhouse imx 2.1.0
| * jailhouse-imx: Keep make install from running pipRodrigo M. Duarte6 days1-1/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | python3-pip-native now ends up in the recipe sysroot, so the jailhouse Makefile finds a usable pip and its install target runs "pip install .". pip then tries to fetch setuptools from PyPI for an isolated build, and do_install fails because the task has no network access: ERROR: Could not find a version that satisfies the requirement setuptools>=40.8.0 (from versions: none) pyjailhouse is already installed by setuptools3_do_install, so that pip call has nothing to add. Disabling pip detection is not an option, because the same PYTHON_PIP_USABLE check also gates installing the Python helpers into libexec (jailhouse-cell-linux, jailhouse-config-create, ...). Pass PIP=true to the install call instead: the check still passes, the helpers are installed, and the pip step becomes a no-op. Tested on imx8mq-evk (NXP BSP, linux-fslc-imx 6.18.24) with core-image-weston: booted with jh_mmcboot, enabled the imx8mq root cell, created, loaded and started the gic-demo inmate, then shut it down, destroyed it and disabled jailhouse. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
| * jailhouse-imx: Upgrade to NXP lf-6.18.37_2.1.0Rodrigo M. Duarte6 days1-2/+2
|/ | | | | | | | | | | | | Move to the lf-6.18.37_2.1.0 branch. The new SRCREV is two commits ahead of the previous one: i.MX937 cell configurations and an SBOM update. No driver or hypervisor code changes. The NXP bbappend for this release also adds a revert of the bin_attribute version check (6.16 -> 6.15). It is not carried here: every kernel we build against is 6.18 or newer, so both checks give the same result. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* Merge pull request #2914 from mdrodrigo/linux-fslc-7.2Otavio Salvador6 days3-5/+45
|\ | | | | Linux fslc 7.2
| * linux-fslc: 6.12.34 -> 7.2.13Rodrigo M. Duarte6 days1-5/+5
| | | | | | | | | | | | | | | | | | | | | | | | Move the recipe to the 7.2.x+fslc branch, at the v7.2.13 tag (commit 3c172eede0d6). Tested by building core-image-weston and booting it on the imx8mq-evk and an i.MX6 DualLite SABRE SD board. etnaviv comes up (GC7000 on imx8mq, GC880/GC320 on imx6) and Weston runs on DSI-1 and LVDS-1. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
| * kernel-module-nxp-wlan: Fix build with Linux 7.2Rodrigo M. Duarte6 days2-0/+40
|/ | | | | | | | | | | | | | | | | The driver uses the probe_peer cfg80211 op on 7.2 kernels, but the rename from probe_client landed in Linux v7.3-rc1, so the build fails against mainline-based 7.2 kernels such as linux-fslc 7.2.13: moal_sta_cfg80211.c: error: 'struct cfg80211_ops' has no member named 'probe_peer' Add a patch moving the version check to 7.3.0. Kernels older than 7.2 (linux-imx, linux-fslc-imx and linux-fslc-lts 6.18) are not affected. Tested by building core-image-weston with linux-fslc 7.2.13 for imx8mq-evk and imx6qdlsabresd; moal and mlan load on imx6qdlsabresd. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* Merge pull request #2913 from mdrodrigo/linux-fslc-lts-6.18Otavio Salvador6 days2-44/+3
|\ | | | | linux-fslc-lts: Update to 6.18.59
| * linux-fslc-lts: Update to 6.18.59Rodrigo M. Duarte6 days2-44/+3
|/ | | | | | | | | | | | | Move the LTS recipe from the 6.6.x+fslc branch to 6.18.x+fslc, as was done for the 6.1 -> 6.6 update, and drop the 6.1 recipe. The 6.18.x+fslc branch is at 6.18.59 (commit 1253a6aa5d10). Tested by building core-image-weston for imx6qdlsabresd (mainline BSP) and booting it on an i.MX6 DualLite SABRE SD board: etnaviv (GC880/GC320) and imx-drm come up and Weston runs on LVDS-1. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* Auto-update LICENSE file with current recipe licensesgithub-actions[bot]7 days1-1/+1
|
* Merge pull request #2912 from mdrodrigo/u-boot-fslc-2026.10Otavio Salvador7 days6-4/+363
|\ | | | | u-boot-fslc: Update to 2026.10
| * u-boot-fslc: Update to 2026.10Rodrigo M. Duarte7 days6-4/+363
|/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The U-Boot fork has been updated to include the tagged v2026.10 from upstream. The following commits were ported from 2025.01+fslc: ---- 322904a1db ("imx: iMX6: Define disconnect_from_pc function") ebf892284f ("MLK-21854-2 imx: uuu: Update autoboot to run uuu or fastboot") 32b6811094 ("pico-imx6ul, pico-imx7d: Explain how to use eMMC user partition") eec398606f ("mx6sabre_common: Dynamically change the video output") 8852a5a1c7 ("mx6sabresd: Enable video interfaces in bootargs") ---- The mkimage signing code still uses the OpenSSL ENGINE API, removed in OpenSSL 4.0, so building the host tools fails. Carry the OpenSSL Provider API patch that OE-Core uses for its own u-boot recipe; it applies cleanly to this tree. It goes in the common include, since u-boot-fslc-mxsboot builds the same tools. Two changes in 2026.10 affect the i.MX 8M boot container that binman builds for the mainline BSP: - fd1072d0d59 ("imx8m: dts: Update ddr firmware name") packs the versioned lpddr4_pmu_train_*_202006.bin files on i.MX 8M Mini, Nano and Quad, and binman fails with missing external blobs otherwise. Select them for the mainline BSP on imx8mq-evk and imx8mm-lpddr4-evk, the LPDDR4 machines that use u-boot-fslc there. The NXP BSP and its imx-boot keep the unversioned files. - 5b6dba2924b ("imx8mq: Correct signed_hdmi firmware position") puts the signed HDMI firmware, with its 1KB head padding, at the start of flash.bin, which must now be written at 32KB. At 33KB the boot ROM does not find the IVT and the board stays silent. Set IMX_BOOT_SEEK to 32 for the mainline BSP on imx8mq-evk. The pylibfdt SWIG 4.5 fix and the binman pkg_resources fix needed by 2025.01 are already part of 2026.10. Tested with 'bitbake core-image-weston u-boot-fslc-mxsboot-native' for imx8mq-evk with the mainline BSP. On an imx8mq-evk board the image boots through U-Boot 2026.10 and Linux to Weston, which runs on the DSI panel with the GC7000 GPU. That boot needed imx-atf built with LOG_LEVEL=40: the release BL31 from imx-atf hangs after its banner on this board, independent of the U-Boot version. imx8mm-lpddr4-evk was not built. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* Merge pull request #2907 from matheusmbar/add_imx952Otavio Salvador7 days10-5/+270
|\ | | | | Add i.MX 952 SoC and its first machines
| * imx952-15x15-lpddr5-frdm: Add machine conf from meta-imxMatheus Barbosa8 days1-0/+65
| | | | | | | | | | | | Based on LF6.18.37_2.1.0 NXP BSP release. Signed-off-by: Matheus Barbosa <matheus.barbosa@toradex.com>
| * imx952-19x19-lpddr5-evk: add machine conf from meta-imxMatheus Barbosa8 days2-0/+168
| | | | | | | | | | | | Based on LF6.18.37_2.1.0 NXP BSP release. Signed-off-by: Matheus Barbosa <matheus.barbosa@toradex.com>
| * imx-oei, imx-system-manager: Add i.MX 952 to COMPATIBLE_MACHINEMatheus Barbosa8 days2-2/+2
| | | | | | | | | | | | Based on LF6.18.37_2.1.0 NXP BSP release. Signed-off-by: Matheus Barbosa <matheus.barbosa@toradex.com>
| * imx-base: Add i.MX 952 configurationMatheus Barbosa8 days2-0/+9
| | | | | | | | | | | | Based on LF6.18.37_2.1.0 NXP BSP release. Signed-off-by: Matheus Barbosa <matheus.barbosa@toradex.com>
| * imx-boot: Add i.MX 952 supportMatheus Barbosa8 days1-0/+20
| | | | | | | | | | | | Based on LF6.18.37_2.1.0 NXP BSP release. Signed-off-by: Matheus Barbosa <matheus.barbosa@toradex.com>
| * optee-os-common-fslc: Add the i.MX 952 platformMatheus Barbosa8 days1-0/+1
| | | | | | | | | | | | Based on LF6.18.37_2.1.0 NXP BSP release. Signed-off-by: Matheus Barbosa <matheus.barbosa@toradex.com>
| * use-imx-security-controller-firmware: Add i.MX 952 configurationMatheus Barbosa8 days1-0/+1
| | | | | | | | | | | | | | | | Add the SECO FW file name for i.MX 952. Based on LF6.18.37_2.1.0 NXP BSP release. Signed-off-by: Matheus Barbosa <matheus.barbosa@toradex.com>
| * imx-base: Fix variables ordering, remove trailing spaceMatheus Barbosa8 days1-3/+4
| | | | | | | | | | | | Fixes: 3e08ba16f9e5 ("imx-base: Add i.MX 95 configuration") Fixes: 7cf349d287a9 ("imx-oei: provide virtual/imx-oei") Signed-off-by: Matheus Barbosa <matheus.barbosa@toradex.com>
* | Merge pull request #2911 from mdrodrigo/imx-atf-clean-buildOtavio Salvador7 days1-1/+13
|\ \ | |/ |/| imx-atf: Build out of tree and from a clean directory
| * imx-atf: Build out of tree and from a clean directoryRodrigo M. Duarte7 days1-1/+13
|/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | imx-atf built in the source tree, into ${S}/build, and do_compile never cleaned it. The TF-A Makefile does not track build options such as SPD or LOG_LEVEL, so after a PACKAGECONFIG change make finds every object up to date and relinks nothing. do_unpack does not depend on PACKAGECONFIG, so it does not rerun either, and do_deploy then ships the bl31.bin from the previous configuration. On imx8mq-evk this showed up after switching from the NXP BSP to the mainline BSP, which drops the optee PACKAGECONFIG. The deployed BL31 was still the SPD=opteed build. Its imx8mq setup sets the BL32 entry to BL32_BASE unconditionally, and the mainline boot container carries no OP-TEE, so the board hung right after the "BL31: Built" banner. The stale binary was reproduced byte for byte by building with SPD=opteed and then without it in the same build directory. Build into ${WORKDIR}/build through BUILD_BASE, clean it before every compile, and deploy from there. The Makefile stays in ${S}, so do_compile runs make with -C ${S}; the default do_compile would skip make, as ${B} has no Makefile. Tested on imx8mq-evk with the mainline BSP by building imx-atf with the optee PACKAGECONFIG and then without it. Each build recompiles everything, the optee bl31 carries the opteed service and the plain one does not. With u-boot-fslc 2026.10, core-image-weston then boots on an imx8mq-evk board through BL31, U-Boot and Linux to Weston. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* Merge pull request #2908 from dolcini/dev/sm-firmware-name-default-scopeOtavio Salvador8 days1-1/+4
|\ | | | | imx-base: Default the System Manager firmware name on i.MX 943/95 only
| * imx-base: Default the System Manager firmware name on i.MX 943/95 onlyFrancesco Dolcini8 days1-1/+4
|/ | | | | | | | | | | The default System Manager firmware name expands to "-" on SoCs without a System Manager, so imx-boot-container no longer skips the copy and the U-Boot build fails on all i.MX 8M machines: cp: cannot stat '.../deploy/images/verdin-imx8mm/-.bin': No such file or directory Fixes: 87e55bf687b4 ("imx-base: Set a default System Manager firmware name") Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
* Merge pull request #2906 from mdrodrigo/fix-libcamera-preferred-versionOtavio Salvador8 days1-4/+4
|\ | | | | imx-base: Prefer libcamera 0.7.2.imx
| * imx-base: Prefer libcamera 0.7.2.imxRodrigo M. Duarte8 days1-4/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The libcamera fork was updated to 0.7.2.imx and the 0.7.1.imx recipe was removed, but the preferred version still pointed to it, so every mx8mm, mx8mq, mx8ulp and mx95 build warned: WARNING: preferred version 0.7.1.imx of libcamera not available Point it at 0.7.2.imx, matching meta-imx rel_imx_6.18.37_2.1.0. Tested with bitbake -e libcamera for imx8mq-evk and imx95-19x19-verdin: the warning is gone and libcamera_0.7.2.imx.bb is selected. Signed-off-by: Rodrigo M. Duarte <rodrigo.duarte@ossystems.com.br>
* | Merge pull request #2905 from mdrodrigo/port-opencv-2.1.0Otavio Salvador8 days4-4/+215
|\ \ | |/ |/| Port opencv 2.1.0
generated by cgit v1.2.3-54-g00ecf (git 2.39.0) at 2026-11-05 04:01:40 +0000