Ting Liu [Mon, 20 Apr 2020 02:32:41 +0000 (04:32 +0200)]
pktgen-dpdk: upgrade to 19.12.0
Modifications:
- 19.12.0 - Add per queue stats when the hardware does not support it. Use
the 'page stats' command to see data and use 'port X' to switch
ports.
- 19.10.0 - Fix up the save/load commands to get correct as in GRE key
command
Fix up the problem with pcap files not getting sent if the
number of packets was small
fix up a crash if the scrn pointer was invalid on startup
fixed the log output in startup
Fixed the save/load scripts to use the correct commands.
Fix the IPv4/6 checksum calculations
- 19.08.0 - Fix linking of Lua library when no pkg-config file is found and
linking liblua.a
Fixed and issue with packet rate not being changed when the packet
size is changed
Change version numering to year.month.patch format
Change lua pktgen.set_mac() to have three args set_mac(<portlist>,
'src|dst', mac_addr) instead of two args
add support for setting via lua src and dst mac addresses
fix up the RSS port configuration options
updated code to adjust the packet tx interval based on command
changes on the packet size and other areas.
- 3.7.1 - Add TTL support to single and range modes. Better docs for dump packets
- Pktgen 3.7.1 will build with DPDK 18.02 to 19.08-rc2, but 18.08 has
a problem with vhost.h VRING_EVENT_F_AVAIL not defined and appears
the Ubuntu 19.04 version I am running does not define it correctly.
- fixed the RX side using TX count for rx_burst command.
- General cleanup
- 3.7.0 - Fixed build issues with DPDK 19.08 as DPDK renamed a lot of defines
Fixed up the meson files to build pktgen with meson and ninja
Minor cleanup
linux-fslc-imx: introduce 5.4.y kernel with NXP patches
Introduce new kernel recipe, which incorporates all patches from
original NXP release 5.4.3-1.0.0 and latest patch level from stable
5.4.y kernel tree from korg.
This kernel recipe is a consolidation of modifications from upstream and
NXP and is meant to be supported by Community with further stable
patchlevel updates, individual patches which pending in upstream to be
included in the stable, and also to receive updates from NXP internal source
tree from the [lf-5.4.y] branch, which corresponds to release 5.4.3-1.0.0.
linux-imx-[mfgtool]-5.4.3: update recipes to align with kernel-yocto
When kernel-yocto class is inherited by kernel recipes, the way how PV
is constructed changes to reflect GIT kernel tags and meta information.
This causes the fetcher search to fail, since paths used to locate files
are compiled from PV.
Replace version-specific paths in recipes to use PN instead, which is
registered in the fetcher as a valid search location.
Linux kernel recipe should have LINUX_VERSION also defined, since it is
used by kernel-yocto class to derive PV. This version is set to the real
kernel version provided by NXP in their source tree.
Inherit the class in order to take advantage of providing mechanisms to
use kernel config fragments, patches and verification features
implemented upstream.
Tom Hochstein [Tue, 5 May 2020 17:16:01 +0000 (12:16 -0500)]
weston: Restore partial clients for i.MX 6 and 7
The clients configuration for fbdev was recently removed from
meta-freescale [1] because of the following error:
```
| Run-time dependency gbm found: NO (tried pkgconfig and cmake)
|
| clients/meson.build:134:4: ERROR: Problem encountered: weston-simple-dmabuf-egl requires gbm which was not found. If you rather not build this, drop "dmabuf-egl" from simple-clients option.
```
It was also removed for non-Wayland from OE-core [2] because of the
following error in the same client dmabuf-egl:
```
clients/meson.build:141:4: ERROR: Problem encountered: weston-simple-dmabuf-egl
requires option renderer-gl which is not enabled. If you rather not build this,
drop "dmabuf-egl" from simple-clients option.
```
linux-qoriq: add perf patch to resolve linux-qoriq build
Kernel 5.4.y for QorIQ platform requires the patch in order to compile
perf component successfully.
Patch has a proper Upstream-Status and points to commit SHA in upstream
kernel tree. It is identical to the one present in the linux-imx recipe,
since kernle tree for i.MX platform also requires it.
optee-os: Replace /lib by ${nonarch_base_libdir} in do_install
This fixes a do_package error when using `usrmerge` in
`DISTRO_FEATURES`. In that case the bin files should have been installed
in /usr/lib. Using ${nonarch_base_libdir} makes the location specific to
distro settings.
Signed-off-by: antznin <agodard@witekio.com> Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Change-Id: I9d6c015975b093f74f29927a7e8bdfd87fd3dc09
The qtbase configure fail to find any suitable GL library, so it fails
with:
,----
| WARNING: No QPA platform plugin enabled! This will produce a Qt that
| cannot run GUI applications. See "Platform backends" in the output of
| --help.
|
| ERROR: Feature 'opengl-desktop' was enabled, but the pre-condition
| '(config.win32 && !config.winrt && !features.opengles2
| && (config.msvc || libs.opengl))
| || (!config.watchos && !config.win32 && !config.wasm && libs.opengl)' failed.
|
| ERROR: Feature 'eglfs' was enabled, but the pre-condition
| '!config.android && !config.darwin && !config.win32 && !config.wasm
| && features.egl' failed.
`----
Fixed: #378
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Ming Liu [Sat, 2 May 2020 11:09:17 +0000 (13:09 +0200)]
imx-boot: fix mkimage_uboot linking issue
In imx-boot's Makefile, it downloads a native tool mkimage_uboot and
executes it, but this native tool requires libssl.so.1.1, which is in
${STAGING_LIBDIR_NATIVE}, we need set LD_LIBRARY_PATH to include it, or
it might run into following errors:
| ./mkimage_uboot -E -p 0x3000 -f u-boot.its u-boot.itb
| ./mkimage_uboot: error while loading shared libraries: libssl.so.1.1: cannot open shared object file: No such file or directory
Currently the kernel module fails to build with following
undefined symbols due to missing Kconfig option.
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:15910:13: error: 'qca_wlan_vendor_ocb_set_config_policy' undeclared here (not in a function); did you mean 'qca_wlan_vendor_setband_policy'?
| 15910 | .policy = qca_wlan_vendor_ocb_set_config_policy,
| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | qca_wlan_vendor_setband_policy
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:15919:13: error: 'qca_wlan_vendor_ocb_set_utc_time_policy' undeclared here (not in a function); did you mean 'qca_wlan_vendor_ll_set_policy'?
| 15919 | .policy = qca_wlan_vendor_ocb_set_utc_time_policy,
| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | qca_wlan_vendor_ll_set_policy
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:15929:13: error: 'qca_wlan_vendor_ocb_start_timing_advert_policy' undeclared here (not in a function); did you mean 'qca_wlan_vendor_attr_ocb_start_timing_advert'?
| 15929 | .policy = qca_wlan_vendor_ocb_start_timing_advert_policy,
| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | qca_wlan_vendor_attr_ocb_start_timing_advert
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:15938:13: error: 'qca_wlan_vendor_ocb_stop_timing_advert_policy' undeclared here (not in a function); did you mean 'qca_wlan_vendor_attr_ocb_stop_timing_advert'?
| 15938 | .policy = qca_wlan_vendor_ocb_stop_timing_advert_policy,
| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | qca_wlan_vendor_attr_ocb_stop_timing_advert
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:15956:13: error: 'qca_wlan_vendor_dcc_get_stats' undeclared here (not in a function); did you mean 'qca_wlan_vendor_attr_dcc_get_stats'?
| 15956 | .policy = qca_wlan_vendor_dcc_get_stats,
| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | qca_wlan_vendor_attr_dcc_get_stats
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:15965:13: error: 'qca_wlan_vendor_dcc_clear_stats' undeclared here (not in a function); did you mean 'qca_wlan_vendor_attr_dcc_clear_stats'?
| 15965 | .policy = qca_wlan_vendor_dcc_clear_stats,
| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | qca_wlan_vendor_attr_dcc_clear_stats
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:15974:13: error: 'qca_wlan_vendor_dcc_update_ndl' undeclared here (not in a function); did you mean 'qca_wlan_vendor_attr_dcc_update_ndl'?
| 15974 | .policy = qca_wlan_vendor_dcc_update_ndl,
| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| | qca_wlan_vendor_attr_dcc_update_ndl
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c: In function '__wlan_hdd_cfg80211_scan':
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:22920:8: warning: ISO C90 forbids variable length array 'chList' [-Wvla]
| 22920 | char chList [(request->n_channels*5)+1];
| | ^~~~
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c: In function '__wlan_hdd_cfg80211_sched_scan_start':
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:28751:9: warning: ISO C90 forbids variable length array 'chList' [-Wvla]
| 28751 | char chList [(request->n_channels*5)+1];
| | ^~~~
| At top level:
| rpb-dunfell/build-rpb-wayland/tmp-rpb_wayland-glibc/work/imx8mqevk-linaro-linux/kernel-module-qca6174/3.0-r0/git/CORE/HDD/src/wlan_hdd_cfg80211.c:25078:20: warning: 'hdd_ieee80211_reason_code_to_str' defined but not used [-Wunused-function]
| 25078 | static const char *hdd_ieee80211_reason_code_to_str(uint16_t reason)
| | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Signed-off-by: Peter Griffin <peter.griffin@linaro.org>
Tom Hochstein [Tue, 28 Apr 2020 19:21:02 +0000 (14:21 -0500)]
weston: Fix non-GBM build error for i.MX 6 & 7
i.MX 6 and 7 doesn't support GBM, required by weston-simple-dmabuf-egl:
| Run-time dependency gbm found: NO (tried pkgconfig and cmake)
|
| clients/meson.build:134:4: ERROR: Problem encountered: weston-simple-dmabuf-egl requires gbm which was not found. If you rather not build this, drop "dmabuf-egl" from simple-clients option.
Remove the clients for i.MX 6 and 7.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
Tom Hochstein [Tue, 28 Apr 2020 14:49:32 +0000 (09:49 -0500)]
weston: Restructure recipe for ease of maintanance
Structure the recipe into two sections, the top as a verbatim
copy of the upstream recipe that is the basis, and the bottom
for the i.MX customization.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
Tom Hochstein [Thu, 23 Apr 2020 13:44:03 +0000 (08:44 -0500)]
waffle: Fix x11 build without GBM support
For i.MX 6 and 7, building x11_egl failed with the following:
```
In file included from ../waffle-1.6.0/src/waffle/surfaceless_egl/sl_window.c:35:
../waffle-1.6.0/src/waffle/surfaceless_egl/sl_platform.h:30:10: fatal error: gbm.h: No such file or directory
30 | #include <gbm.h>
| ^~~~~~~
```
The failure is not surprising since i.MX 6 and 7 don't
support GBM. What is surprising is surfaceless_egl is
built even if it is disabled. The meson dependency
for surfaceless_egl was incorrectly tied to x11_egl,
so fix that, plus add a dependency on GBM.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
Tom Hochstein [Thu, 23 Apr 2020 13:38:34 +0000 (08:38 -0500)]
waffle: Add missing dependency on EGL
For Wayland builds without XWayland support, the following
build break was happening:
In file included from /home/bamboo/build/5.4-zeus-full/fsl-imx-internal-wayland/temp_build_dir/build_fsl-imx-internal-wayland/tmp/work/aarch64-mx8-poky-linux/waffle/1.6.0-r0/recipe-sysroot/usr/include/EGL/egl.h:39,
from ../waffle-1.6.0/src/waffle/egl/wegl_context.h:31,
from ../waffle-1.6.0/src/waffle/egl/wegl_util.c:28:
/home/bamboo/build/5.4-zeus-full/fsl-imx-internal-wayland/temp_build_dir/build_fsl-imx-internal-wayland/tmp/work/aarch64-mx8-poky-linux/waffle/1.6.0-r0/recipe-sysroot/usr/include/EGL/eglplatform.h:144:10: fatal error: X11/Xlib.h: No such file or directory
144 | #include <X11/Xlib.h>
| ^~~~~~~~~~~~
compilation terminated.
This happens because the egl.pc configuration is missing.
Fix the mising dependency.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
gstreamer1.0-plugins-imx: Update to version 0.13.1
* Last planned release in the 0.x series. Development now focuses on 2.x.
The 2.x series will support i.MX6, i.MX7, and i.MX8.
* OE recipe fix to use the correct G2D dependency (imxg2d, not imx-gpu-viv)
* waf: Update to version 2.0.12 and switch to Python 3
* misc:
* gst_buffer_pool_config_set_allocator() and get_allocator() are
transfer-none. Changes to the code address that, fixing various refcount
related problems.
* Update GstPhysMemoryAllocator support for GStreamer 1.14+
* v4l2video:
* v4l2src: Set interlace flags in GstBuffer
* v4l2sink: Define V4L2_COLORSPACE_DEFAULT if it doesn't exist
* v4l2: Add phys mem meta only if the driver/card is known to support this
This makes USB video cameras work with imxv4l2videosrc, because they do
not provide any physical address.
* v4l2_buffer_pool: First reset the buffer storage in the pool, then queue
the buffer. Otherwise it might happen that we dequeue the same buffer in
a different thread already *before* we have it reset in the buffer storage,
and then run into an assertion because it contains NULL.
* vpu:
* decoder: Fix for an endless loop at gst_imx_vpu_decoder_finish() when in
no_wait mode
* decoder: Clear the GST_VIDEO_BUFFER_FLAG_TFF flag in case of
IMX_VPU_INTERLACING_MODE_BOTTOM_FIELD_FIRST
* decoder: Honour num-additional-framebuffers setting
* decoder: Fix leak when input video formats change due to a missing
gst_imx_vpu_decoder_close() call
* ipu: Reintroduce interlace methods
* audio:
* Save pointer to codec_data GstBuffer to fix memory leak
* Only allocate reordered_channel_positions if needed. If we don't need to
reorder the audio channels, we the unnecessary allocation otherwise.
* g2dpango:
* Add missing header and cleanup a few warnings
* Unref video_frame memory after phys_memory check
Signed-off-by: Carlos Rafael Giani <crg7475@mailbox.org>
Peter Griffin [Wed, 22 Apr 2020 22:52:14 +0000 (23:52 +0100)]
optee-test_3.2.0.imx: fix TA_DEV_KIT_DIR is not correctly defined error
DEFAULTTUNE is no longer aarch64 which means optee-test
build fails. Update to use the same mechanism as optee-os
recipe for setting arch.
Build now passes on imx8mq-evk and optee xtest suite
passes
+-----------------------------------------------------
16078 subtests of which 3 failed
74 test cases of which 3 failed
0 test case was skipped
TEE test application done!
Signed-off-by: Peter Griffin <peter.griffin@linaro.org>
-Update EULA to v10
-Add vpu wrapper unit test for i.MX8
Enable 8mm/8mq to support H264/HEVC byte-stream format decoding
Enable 8mm to support H264 encoding
-Update EULA to v10
-Fix unexpected EOS reported by Aiurdemux while playing fragmented mp4 video
-Add Mpeg2ResetOuputBuffer when track is disabled
-Upgrade mp 3parser's getNextSamle, don't read 16KB
because this mode can't calculate right timestamp
-Add Parser API to get ac4 presentation info
-Fix ID3 parser return wrong artwork size
* imx6: replace mxcfb.h check with i.MX6 specific imx header check
the mcxfb.h check only makes sense with i.MX6 devices, so requiring
those for others like i.MX8 led to build errors
* update waf to 2.0.12 and switch wscript to use Python 3
* imx6: fix encoder pointer usage in vpu_EncGiveCommand() call
Signed-off-by: Carlos Rafael Giani <crg7475@mailbox.org>
imx-base.inc: Allow for setting different IMX_DEFAULT_BSP values
Current assignment of the IMX_DEFAULT_BSP variables is too strong. For
example, any IMX_DEFAULT_BSP value set in local.conf currently gets
overridden by the assingment in imx-base.inc. Soften the assignment
to allow for alternative values for these variables.
Signed-off-by: Carlos Rafael Giani <crg7475@mailbox.org>
Certain packages are using linux-imx-headers, which enforces them to use
NXP BSP since headers are provided from the NXP kernel tree.
Those recipes, which requires NXP kernel headers, are now including the
use-imx-headers class to make sure the correct headers package is used.
Since the introduction of separation between NXP and Mainline BSP for
certain machines, recipes which are NXP-dependent are getting their
COMPATIBLE_HOST to be marked as (null) effectively making them
incompatible with mainline BSP which is expected behavior.
By extending this restriction into the use-imx-headers class, all
recipes which inherits is (because it is required) are getting
automatically restricted to NXP BSP, hence making package NXP-dependent
much easier.
linux-imx-headers is marked as "NXP-only" but it is done explicitly
inside the recipe in order to avoid confusion of inheriting the class.
machine: imx[6,7,8]: update dtb references for 5.4 kernel
Updated kernel has some DTS files added and some dropped for various
i.MX derivatives. Those should be aligned with respective machines in
order to deliver proper DTB files, and do not fail the build for those
are missing.
Device tree files listed in respective machines are those only present
in updated 5.4 kernel release from NXP. Derivatives, which relied on
those device trees that are discarded from the new kernel release should
backport their respective files onto that update.
In other words: if after this update the DTS file used for the HW happens
to be missing - it should be manually added for that HW, since the default
version is no longer provided by NXP in their updated kernel tree.
gstreamer: upgrade to MM_04.05.03_1911_L5.4.0 branch
Upgrade imx fork of gstreamer components to version taken from
MM_04.05.03_1911_L5.4.0 branch. This branch is aligned with the new
kernel update from NXP.
Upgrade weston recipe version 6.0.1 -> 8.0.0 with following
modifications performed:
- Transition build system to meson, replace autotools references with
meson options;
- Align recipe with the latest version from oe-core, include all
configuration changes done upstream into this recipe;
- Drop patches applied upstream and those related to autotools;
linux-imx: upgrade kernel recipe to 5.4.3-1.0.0 from NXP
Upgrade kernel recipe 4.19.35 -> 5.4.3, drop patches which are applied
upstream and re-submit new kernel configuration files for arm32 and
aarch64 build configurations.
Kernel recipe version 4.19.35 is dropped from this layer in favor of the
updated one.
Tom Hochstein [Mon, 13 Apr 2020 16:16:15 +0000 (11:16 -0500)]
imx-gpu-viv: Upgrade to 6.4.0.p2.0
Key Features of the Vivante 6.x Driver Series
- Full support for Khronos OpenGL ES 3.2 API. The new OpenGL ES 3.2 and OpenGL ES Shading Language 3.20 specifications bring AEP,
plus additional functionality, into core OpenGL ES. Vivante OpenGL ES 3.2 implementation has passed Khronos OpenGL ES 3.2 Conformance Test.
And Vivante GC7000XS* GPU core has been certified by Khronos as an OpenGL ES 3.2 compliant product.
- Backward Compatible with 5.x and 4.x Drivers. Existing OpenGL ES 3.x/2.0 applications can run on these drivers without any changes.
6.4.0 P2 GA
======================
Key Features:
- Vulkan 1.1
- OpenVX 1.2
- OpenCL and OpenVX bridged support.
- GPU accelerated ML support
- Upgrade to weston 7.0
- OpenCV 4.0.1
- Chromium v74
- Support for renderdoc 1.2
- vProfiler/vAnalyser through new Vivante IDE
- Support for Android 10
- Support for TFlite 1.3
- NNCTS 1.2
- Optimize command buffer size, so the memory usage per context will be lower
Known Issues :
LF-308 wl_display_connect Failure reported during VPU and GPU stress power management test.
LF-868 [MX6SX-SDB]GPU: weston stopped abnormally during stress suspend resume test.
LF-459 [8QM] GPU/xwld:Vulkan sdk demo 'ObjectSelection' flicker on display.
LF-178 GPU/XWLD: some tests failed in OpenCV performance test.
LF-262 GPU:Some qt5 demos not work. QT5.3 have regressions. Require qt3d update in QT5.4 and later
LF-836 GPU/XWLD: met version mismatch when using renderdoc remote test
LF-861 some tests failed in OpenCV stress test.
LF-423 ML/AI:Some dnn_vxc cases showed out-of-range recognition rates.
LF-844 [6QP_SD] GPU/FB: Bloom_FB cannot be killed by ctrl c randomly.
LF-1109 [8QM] GPU/Vivante IDE tool: vpd file of OpenVX import failed to VSI IDE
LF-1110 [8QM] GPU/Vivante IDE tool: some binary of sample project cannot work after building successfuly in vCompiler test
LF-936 [MX8MQ/MM/MN]GPU:no desktop UI and weston@root.service load fail after run low bus freq test.100%
LF-88 [8QXP_MEK/8QM/8MM_EVK] GPU/XWLD: apitrace tool can not work properly
The 4.14.98-2.3.0 BSP targeting latest SoC silicon revisions uses an even
newer license text 'LA_OPT_NXP_Software_License v10 December 2019'.
And the upcomming 5.4.3-2.0.0 BSP uses yet another one:
'LA_OPT_NXP_Software_License v11 February 2020'.
Add their md5sums additionally.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
