| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
| |
Fixes build errors e.g.
tests/egl/egl-util.c:89:9: error: incompatible integer to pointer conversion returning 'Pixmap' (aka 'unsigned long') from a function with result type 'EGLNativePixmapType' (aka 'struct wl_egl_pixmap *') [-Wint-conversion]
return XCreatePixmap(state->dpy, state->win,
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
|
|
| |
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
|
|
|
| |
The recipe is for all i.MX 6-8 SOCs using NXP BSP, but it confusingly
uses an indirect override imxfbdev|imxgpu. The override used does
happen to cover all 6-8, but it is simpler to just use i.MX series
overrides.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
| |
Used by i.MX 93.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
| |
The NXP repos are moving to github.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
|
|
| |
Includes:
063bd699 MGS-6873 tests/util: add imx-dcnano driver
c6cf6625 MA-20797 Update imx_drm header to enable fence support in DRM blitter API
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
| |
Reset SECURITY_CFLAGS when using clang to fix build
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
|
|
|
|
| |
They do not build for imx8 on master
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
|
|
|
|
| |
glx is not available in imx gpu GL implementation
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Update to the version used in NXP Yocto branch kirkstone-5.15.71-2.2.0.
Updated the repo location to new Github account.
Commit history:
3efd75a MGS-6902 Add SCR for g2d-samples
0094ed8 LF-6616 g2d-samples intermittent build break
80d5888 MGS-3854 Fix global alpha on iMX8QM/QXP
0c36b18 MGS-6862-2 warp_dewarp_test: Fix misuse of &ctx->handle
b4443cb MGS-6862-1 warp_dewarp_test: Add g2d_open check
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
|
|
|
|
|
|
|
|
| |
- Update to the version used in NXP Yocto branch kirkstone-5.15.71-2.2.0
- Also the binaries seem to be stripped now, so skip that step to avoid
build errors
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
|
|
|
|
|
|
|
|
|
|
| |
- Update to the version used in NXP Yocto branch kirkstone-5.15.71-2.2.0
- The 8M Mini does not support OpenCL, so install an SOC-specific G2D
library built without OpenCL. (see e3e0817b in meta-imx)
- Also the binaries seem to be stripped now, so skip that step to avoid
build errors
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
|
|
|
|
|
|
| |
Update to the version used in NXP Yocto branch kirkstone-5.15.71-2.2.0.
Signed-off-by: Gary Bisson <gary.bisson@boundarydevices.com>
|
|
|
|
|
|
|
| |
current implement only includes INI_UNCOMMENT_USE_G2D
to INI_UNCOMMENT_ASSIGNMENTS on the platforms with VIV G2D hardware.
Signed-off-by: Jun Zhu <junzhu@nxp.com>
|
|
|
|
|
|
|
| |
The main recipe now sets xwayland=true in weston.ini, so drop it from
here.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
| |
Update to the version used in lf5.15.52_2.1.0.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Update to the version used in lf5.15.52_2.1.0.
imx-g2d-samples now is configured with the BUILD_IMPLEMENTATION
environment variable which is expected to be set to one of
dpu|gpu-drm|gpu-fbdev depending on the targeted SoC.
Commit history:
* 5a38513 wayland_dmabuf_test:Convert to xdg-shell
* c8167c1 wayland_cf_test:Convert to xdg-shell
* 2ccdebe wayland_shm_test:Convert to xdg-shell
* 82169b8 LF-6057: g2d_wayland_dmabuf_test had no display
* 63272d5 Add multiblit_test build for imx6/7 fbdev
* 85ef72a rename extended_test to tiling_test
* 2fbe894 YOCIMX-6294: Configure samples to build by implementation
* 4f03293 README.md: Cleanup minor issues
* 0b6981e MGS-6648 [#ccc] Add Test loop parameter in command line for power measurements
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
|
|
|
|
|
|
| |
Update to the version used in lf5.15.52_2.1.0.
License was updated to NXP EULA v39 August 2022.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
|
|
|
|
|
|
| |
Update to the version used in lf5.15.52_2.1.0.
License was updated to NXP EULA v39 August 2022.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Update to the version used in lf5.15.52_2.1.0.
Note that the the vulkan installable client driver is provided with a
filename and soname that no longer clashes with the vulkan loader.
Thus the rename and soname magic is dropped.
License was updated to NXP EULA v39 August 2022.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
tests: Add dependency on screenshooter client protocol
Given that the test-helper code relies on the screenshooter protocol,
make sure it's available for us to build, and the dependency ensures we
build in order.
Link: https://github.com/wayland-project/weston
Commit: 2ac6b6b084a877adde64db7faff2ed22eb3ea97a
Fixes: #1273
Signed-off-by: Maxim Paymushkin <maxim.paymushkin@gmail.com>
|
|
|
|
| |
This reverts commit 4d97ada2fa09f83dbb551370deb320e5c74545d3.
|
|
|
|
|
|
|
|
|
|
|
| |
After rebase meta-freescale, there is a building warning as below:
WARNING: imx-gpu-viv-1_6.4.3.p4.2-aarch64-r0 do_package_qa:
QA Issue: imx-gpu-viv: recipe doesn't inherit features_check
Because the features_check was missed when NXP upgraded
imx-gpu-viv package, add it back.
Signed-off-by: Meng Li <Meng.Li@windriver.com>
(cherry picked from commit 14916cfe741f696545515ddc768da43d2ea29fb6)
|
|
|
|
|
|
|
|
|
|
|
|
| |
The logic to move the .so into the main package is incorrect and results
in all libraries being moved to the dev package, causing a QA error like
the following:
```
ERROR: imx-gpu-sdk-5.7.1-r0 do_package_qa: QA Issue: imx-gpu-sdk rdepends on vulkan-loader-dev [dev-deps]
```
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
| |
Add the missing json file for the i.MX vulkan implementation.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
|
| |
The library libvulkan.so is loaded dynamically and so is needed in the
main package.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
| |
When i.MX GPU-based vulkan is used, the mesa drivers are not needed.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
|
| |
i.MX GPU drivers work with an older version of vulkan, so add the older
recipes.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
| |
Signed-off-by: Jose Quaresma <jose.quaresma@foundries.io>
|
|
|
|
|
|
|
| |
With this the libvulkan-imx is recommends only when the machine
supports Vulkan.
Signed-off-by: Jose Quaresma <jose.quaresma@foundries.io>
|
|
|
|
|
|
|
|
|
|
| |
OpenVX is fully supported on i.MX 8 QuadMax and 8 QuadPlus. A recent
change removed OpenVX packages except for this case. This was a mistake,
since there is in fact limited OpenVX support provided on other i.MX 8
machines as needed for i.MX machine learning packages.
Fixes: 09e2ce3
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
|
|
|
|
|
| |
When removing rdp as default dependency a hidden buildtime
dependency on libdrm is uncovered for targets using fbdev
as backend and g2d as renderer. As a result compiling for
i.MX6 will fail due to missing libdrm headers.
Add libdrm to DEPENDS for imxfbdev.
Signed-off-by: Markus Niebel <Markus.Niebel@ew.tq-group.com>
|
|\
| |
| | |
Some fixes for imx-nxp-bsp
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The for loops over the FILES:xxx variables do file expansion on the
FILES entries. As a result any file name(s) specified with a wildcard
which actually exists in the host filesystem will be replaced with
the host filesystems file name(s).
Disable file expansion for the for loop and reenable it afterwards
so that the expansion happens on the full file names with the ${D}
prefix.
E.g if the host has /usr/lib/libOpenCL.so.1, then
/usr/lib/libOpenCL.so.* gets expanded to /usr/lib/libOpenCL.so.1,
and as a result ${D}/usr/lib/libOpenCL.so.1.2, ${D}//usr/lib/libOpenCL.so.3 ...
will not be deleted.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
imx-gpu-g2d dynamically links against the library so OE stops with
the following errors:
| ERROR: imx-gpu-g2d-6.4.3.p4.2-r0 do_package_qa: QA Issue: /usr/lib/libg2d-viv.so.2.1.0 contained in package imx-gpu-g2d requires libOpenCL.so.3(OPENCL_1.0)(64bit), but no providers found in RDEPENDS:imx-gpu-g2d? [file-rdeps]
| ERROR: imx-gpu-g2d-6.4.3.p4.2-r0 do_package_qa: QA Issue: /usr/lib/libg2d-viv.so.2.1.0 contained in package imx-gpu-g2d requires libOpenCL.so.3()(64bit), but no providers found in RDEPENDS:imx-gpu-g2d? [file-rdeps]
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Previous commit did not really work because we are using bb.utils.filter
and that removed the artificial space sadly. Therefore add it outside
this function call.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
| |
| |
| |
| |
| |
| |
| | |
This can cause problem by creating unknown packageconfigs by
concatenating the strings
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The weston_10.0.0.imx now depends on freerdp which in turn pulls
some unrelated big dependencies like cups and gstreamer.
| # meta-freescale/recipes-graphics/wayland/weston_10.0.0.imx.bb
| PACKAGECONFIG[rdp] = "-Dbackend-rdp=true,-Dbackend-rdp=false,freerdp"
|
| PACKAGECONFIG:append = " \
| rdp \
| ${@bb.utils.filter('DISTRO_FEATURES', '${PACKAGECONFIG_OPENGL}', d)}"
| # meta-openembedded/meta-oe/recipes-support/freerdp/freerdp_2.8.0.bb
| PACKAGECONFIG ??= " \
| ${@bb.utils.filter('DISTRO_FEATURES', 'directfb pam pulseaudio wayland x11', d)}\
| gstreamer cups pcsc \
| "
|
| PACKAGECONFIG[cups] = "-DWITH_CUPS=ON,-DWITH_CUPS=OFF,cups"
| PACKAGECONFIG[gstreamer] = "-DWITH_GSTREAMER_1_0=ON,-DWITH_GSTREAMER_1_0=OFF,gstreamer1.0 gstreamer1.0-plugins-base"
Signed-off-by: Jose Quaresma <jose.quaresma@foundries.io>
|
|
|
|
|
|
|
|
| |
OpenCL and OpenVX components are provided for SOCs that are not
supported. Fix this by removing them from the PACKAGES variable and
from the install.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
|
| |
The FILES variable should always specify the components regardless of
whether those components work for the SOC.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|
|
|
|
|
|
|
| |
The removal of the libvdk-imx and libvulkan-imx components is properly
handled with the existing FILES variables instead of with an incomplete
hard-coded list.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
|\
| |
| | |
imx-gpu-viv: Add support for i.MX 8 without Wayland
|
| |
| |
| |
| |
| |
| | |
This configuration is experimental and has never been fully tested.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
For the i.MX 8M Mini the GPU doesn't provide the Vulkan API and thus
the Vulkan libs are not installed. Delete them in that case from the
install directory to prevent the following build time error:
| ERROR: imx-gpu-viv-1_6.4.3.p4.2-aarch64-r0 do_package: QA Issue: imx-gpu-viv: Files/directories were installed but not shipped in any package:
| /usr/lib/libvulkan.so.1.2.1
| /usr/lib/libvulkan.so.1
Once the above files are deleted from ${D} I additionally get:
| ERROR: imx-gpu-viv-1_6.4.3.p4.2-aarch64-r0 do_package_qa: QA Issue: -dev package imx-gpu-viv-dev contains non-symlink .so '/usr/lib/libSPIRV_viv.so' [dev-elf]
Why it is not in the list of the `installed but not shipped`files
I do not know, but as libSPIRV_viv is also packaged into the vulkan
packages let's delete it too.
Fixes: ee92b5a8 ("imx-gpu-viv: Mark libvulkan-imx unavailable for 8M Mini")
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|\ \
| | |
| | | |
musl fixes
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
These libs are pre-compiled for glibc therefore, on musl they will need
redirecting via the gcompat stub to work at runtime.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
These are glibc specific symbols which gcompat should be able to handle
at runtime
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
| | |
| | |
| | |
| | | |
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
| | |
| | |
| | |
| | | |
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This ensures that the symbols from the prebuilt libs are not flagged
during build, we are betting on gcompat to provide the missing glibc
API stubs at runtime.
Depend on gcompat on musl needed for imxgpu machines its needed
since graphics driver is precompiled for glibc.
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|