While we are working to update our platforms to support Weston 10 and
beyond, we are falling back to software rendering for platforms that do
not currently have a solution for GPU drivers (namely the boards that
use SGX).
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Set SGX display controller alias variables accordingly, as SGX has a few
different aliases across platforms.
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Remove default preferences for virtual/egl, virtual/libgles and virtual/libgbm
from machine configs to follow the new dependency chains via mesa in oder to
support accelerated and software-rendered graphics, which is selected by
virtual/gpudriver preference.
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Patched mesa will now be configured based on what virtual/gpudriver
preferred provider is selected. The gpu drivers will now runtime
recommend the associated UM components.
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Locally overlay verbatim copies of mesa 22.0 patches from oe-core/kirkstone,
so the same codebase works in kirkstone and master. Since upstream version
of mesa in master is now 22.3 and no longer carries these patches, but they
are still needed for 22.0-based PVR fork of mesa used with Rogue driver.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
While we are working to update our platforms to support Weston 10 and
beyond, we are falling back to software rendering for platforms that do
not currently have a solution for GPU drivers (namely the boards that
use SGX).
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Attempt to update the SGX display controller alias variable according to
what Andrew told me. SGX apparently had a few different aliases across
platforms.
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Remove all gpu machine features and instead rely on the new dependency
chain for resolving gpu dependencies. Set preferred providers for
virtual/gpudriver where needed.
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Patched mesa will now be configured based on what virtual/gpudriver
preferred provider is selected. The gpu drivers will now runtime
recommend the associated um components.
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Original READMEs with bootflows were reworked and moved to a dedicated doc/
folder in more recent U-boot versions. Update links in the k3r5 config file
accordingly to provide some basic explanation of K3 bootflows.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
UIO drivers are generally frowned upon in the community, and for PRU
we now have much better alternatives in RProc, RPmsg, and PRU-Consumer.
Let's not send the wrong message by continuing to support this crufty
interface. Remove the kernel module, DTBs, and test recipe.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
BeagleBone AI-64 (https://beagleboard.org/ai-64) uses Texas Instruments
Jacinto TDA4VM/J721e SoC. Officially BeagleBone AI-64 supports Debian
Linux and builds its BSP on top of meta-ti and TI SDK, but adds custom
DTBs and DTBO overlays, plus integrates other drivers and features.
Let's add corresponding recipes for beagleboard.org kernel and u-boot
along with the BeagleBone AI-64 machine config.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
The ti-sgx-ddk driver requires an additional userspace initialization
step after the kernel module has probed the device. Without this
initialization, no EGL context can be created and Weston etc. will fail to
start.
The driver package contains an init script, but this does not work on pure
systemd-based systems without sysvinit compat support (as provided by the
systemd-compat-units package). Introduce an enabled-by-default
PACKAGECONFIG that installs a udev rule instead to run the init command
automatically when the driver is loaded, solving the issue without
depending on a specific init system. On builds without udev, this
PACKAGECONFIG can be disabled, reverting to the old init script-based
solution.
udev reports several events when the pvrsrvkm module is loaded:
- add event for the kernel module
- add events for two DRM devices
- bind event for the GPU platform device
The DRM devices aren't nice to match on, and the kernel module add is
too early to run `pvrsrvctl --start`, so we trigger on the platform
device bind.
Tested with Weston 9.0.0 on the AM65x-based TQ-Systems MBa65xx.
Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
These URIs have been updated in dunfell, but two instances were
forgotten in kirkstone/master.
Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
The TI_K3_SECDEV_INSTALL_DIR is defined on the meta-ti-extras layer
so when the layer is not used the variable is undefined.
For such cases we can use default value that is the same defined
on the recipes-ti/includes/ti-paths.inc
Signed-off-by: Jose Quaresma <jose.quaresma@foundries.io>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
The TI_K3_SECDEV_INSTALL_DIR_RECIPE is defined on the meta-ti-extras layer
so when the layer is not used the variable is undefined.
For such cases we can use default value that is the same defined
on the recipes-ti/includes/ti-paths.inc
Signed-off-by: Jose Quaresma <jose.quaresma@foundries.io>
Acked-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
The meta-ti-bsp layer is still broken for K3 HS platforms.
The patch [1] makes the check-layer happy again as it does not
fail any more during bitbake parsing but it will be difficult
for the end user to discover this dependency on meta-ti-extras.
On the other side, meta-ti-extras depends on meta-ti-bsp so
this is a circular dependency.
Currently the layer is broken for am62xx-evm/am64xx-evm machines:
| ERROR: Nothing PROVIDES 'ti-k3-secdev-native' (but mc:k3r5:/srv/oe/build/conf/../../layers/meta-ti/meta-ti-bsp/recipes-bsp/ti-sci-fw/ti-sci-fw_git.bb DEPENDS on or otherwise requires it). Close matches:
| libtspi-dev-native
| makedevs-native
[1] - 8e43835c ti-sci-fw: make dependency on meta-ti-extras soft
Signed-off-by: Jose Quaresma <jose.quaresma@foundries.io>
Acked-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
The WIC image for K3 depends on the TI-SCI firmware, on AM65x and J721e
also the bootloader. Make this clear by only depending on firmware in
the base k3.inc file, then adding the bootloader dependency in AM65x and
J721e specific includes.
Remove unneeded overrides from the evm level files.
Remove do_image_tar dependencies as the tar files do not include
the bootloader nor TI-SCI firmware, only the boot partition of
the WIC file needs these.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
These defines are specific to a SoC, not a SoC family (K3) nor
any specific board. Move these to the SoC level for all boards
to be consistent with boards already doing this correctly.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
This was used in the base Linaro optee-os_git.bb recipe, it is
no longer used, remove it from our boards.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
J7 is an unneeded level of grouping that is now causing issues and
leading to mistakes when adding new J7 based platforms. Each J7x
family device should use the SoC name specifically to prevent
accidentally adding or removing features that are not meant
for the new SoC.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Now that we have SoC names, we can avoid adding features based on the
board name. We expect folks to create their own boards based on these
SoCs, and so using the TI made EVM board name everywhere adds extra churn
when adding a new board. Plus it is more correct for most of these
features as they depend on the SoC, not on the EVM board.
One other thing we do here is to not use the generic "j7" name,
the current and future J7 devices are far to feature diverse
to group at this level. Grouping like that will lead to the wrong
things getting enabled as new J7 SoCs are added.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Like we do with J721e, J7200, and J721s2, use a SoC header include file
to define common properties. This eases adding new boards based on these
SoC.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Multi-cert is already now the default, no need to add a comment on that.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Upstream meta-arm has updated optee-os to 3.19 and our local changes
conflict with that. Update bbappend to match upstream.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
These were in the meta-arago layer, but any distro should be able
to run on HS devices. So move the classes and setup to the BSP layer.
Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
In many spots we have to use the full EVM name for each EVM using the
AM57x SoC leading to missed features on some EVMs. Add the am57xx name
so we can generically add for AM57x.
Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Rather than just support mickledore we should also support the other
layers that we still support to ensure that as many as different
versions we can.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Move to setting the values for PREFERRED_PROVIDER using the
?= default assignment so that we can override the setting if
we would like to.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
There were fuzz errors with the patch that needed to be cleaned up,
and for some reason a v2 that was supposed to move the duplicate
patches in files/ was not pushed properly before merging.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
These bbappends were bumping versions to 3.17 when upstream only provided
3.16 version. Now that upstream meta-arm provides 3.18 versions, there's
no need for the bbappends, especially since they break in do_install.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Replace "require" with "include" to load ti-paths.inc and not break
parsing when meta-ti-extras is not in the BBLAYERS list.
As ti-paths.inc is only needed to define TI_K3_SECDEV_INSTALL_DIR
for ti-k3-secdev when TI_SECURE_DEV_PKG_K3 is not set and building
for K3 HS platforms. Since ti-k3-secdev resides in meta-ti-extras,
the soft dependency will still work fine and give a build error
"Nothing PROVIDES 'ti-k3-secdev-native'" when it is needed.
Also, since TI_K3_SECDEV_INSTALL_DIR can now be undefined, use
d.getVar() to access it ensuring it doesn't break the condition.
Reported-by: Daniel Díaz <mrchapp@gmail.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
These patches are from upstream against v3.18, but we moved to v3.19 and the
patches are now invalid. These can be removed once upstream moves up to or
beyond v3.19.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Cleanup and unify again - omapl138-lcdk is the only platform supported in
the OMAP-L1 family. Plus there's no longer DaVinci family with corresponding
settings. Adjust accordingly.
Please note, there's currently a bug building Rust on ARMv5 platforms:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=14967
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
CMEM has been deprecated for removal for some time now. All users should
have migrated to the Linux upstream DMA-BUF Heaps framework. Remove CMEM.
We only remove the kernel module part in this patch to prevent builds
from breaking in the meta-arago layer due to dependencies on the CMEM
library. These packages should either be updated or disabled. After
that we will remove the library.
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Update to the latest upstream master which is 3.19.0.
Signed-off-by: Praneeth Bajjuri <praneeth@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Bump the source revision for the rogue pvr kernel module.
Source changelog:
- A hack was added to increase stability on J7 platforms while we are
looking into a proper fix for cache coherency issues in this module
Signed-off-by: Randolph Sapp <rs@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Otherwise undefined variable is returned literally as ${TI_SECURE_DEV_PKG_K3}
and the condition won't work.
Cc: Andrew Davis <afd@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Include ti-paths.inc so TI_K3_SECDEV_INSTALL_DIR_RECIPE gets defined and
artifacts are installed in the correct location.
Cc: Andrew Davis <afd@ti.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Ryan Eatmon
Randolph Sapp
Denys Dmytriyenko
Andrew Davis
Matthias Schiffer
Jose Quaresma
Khem Raj
Praneeth Bajjuri