This is no longer used anywhere, drop it out.
Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
If you are not building a TI machine then KERNEL_BASE_CLASS is not
defined and you get a parse error. Move the default setting out of
ti-soc.inc and into ti-kernel.inc.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
The legacy HS platforms (am335x-hs-evm, am437x-hs-evm, and
am57xx-hs-evm) are not using the new fitImage flow, but the kernels all
include the kernel.bbclass which disallows the old fitImage flow.
Copy in the last version of the kernel.bbclass and name it
kernel_legacyhs.bbclass, and update the machine configs to defined a
variable to use as the KERNEL_BASE_CLASS to inherit.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Switch all kernel recipes and K3 based platforms to use the new
FIT image wrapper functionality.
Update FIT default DTB entry to not contain vendor dubdir, as it
is now handled through a conf prefix.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
The kernel class operates a little differently. The kernel is
downloaded and then moved to the "global" shared location and a symlink
is created to point to that location. Since we cannot set S =
"${WORKDIR}/git" anymore, just choose a new name under UNPACKDIR. The
value for BB_GIT_DEFAULT_DESTSUFFIX is the subdir name that is used by
default when fetching git repositories, so just use that name.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
A recent set of changes on oe-core has dramatically revamped the
fitimage generation code. It is initially unclear how to fold those
change into our flow, so in the short term we will copy the previous
versions of the kernel bbclass files from a commit before [1] any
changes to the class files.
As soon as we figure it out, these files will be removed. This just
help unblock anyone trying to build on master including meta-ti-bsp.
[1] https://git.openembedded.org/openembedded-core/commit/?id=fdb0a51598156f99aa91f7495d7eada92a459e97
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
The script that sends the automated patch does not currently handle
updating the KERNEL_REPRODUCIBILITY_PATCHES variable with the changes in
the patches.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
It was suggested to move the reproducibility patches out of
ti-kernel.inc and into the respective recipes.
Taking this time to also rework all of the pieces that go into the SRC_URI
and create a standardization for how all of the kernel recipes are done.
Create KERNEL_GIT_URI, KERNEL_GIT_PROTOCOL, KERNEL_GIT_BRANCH,
KERNEL_DEFCONFIG, and KERNEL_REPRODUCIBILITY_PATCHES. The default for
the above new variables is to the git.ti.com repository and how we manage
the kernel settings. The other kernel recipes that do something other than
how TI does it can simply override the variables.
This change will make it easier to perform the upstream testing that we
do on a nightly basis. Additionally, it will simplify the auto generation
of the patches to move the mainline (stable kernel) recipes since the
required patches will be all done in the same recipe.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Share the most commonality between the recipes, except:
* SRC_URI torvalds vs. linux-next
* SRCREV and PV
* Device Tree list vs. DT prefix
* SUMMARY
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
MACHINE_KERNEL_PR is a remnant of the early OE days and hasn't been
needed for quite some time. Let's finally remove it completely.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Instead of making kernel RDEPENDS on devicetree and different FW images with
lots of machine-override appends that are hard to modify downstream, use proper
MACHINE_ESSENTIAL_EXTRA_RDEPENDS and MACHINE_ESSENTIAL_EXTRA_RRECOMMENDS vars
that are specifically designed for this purpose.
This also makes such dependencies generic and not tied to a particular kernel
recipe including kernel-rdepends.inc with the entire machine-specific list.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Acked-by: Andrew Davis <afd@ti.com>
Acked-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Mainline should test the mainline kernel as it is, not with out-of-tree
changes like new DTB/DTBOs. Use only upstream DTB/DTBOs. As that was the
last thing provided by ti-upstream-tools, remove it here.
Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Mainline should test the mainline kernel as it is, not with out-of-tree
changes like to the default configuration that one would get when using
a stock distro. Use only upstream configuration.
Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
We often carry more dtbs/dtbos in our kernel than we have upstreamed.
The inclusion of all of the dtb/dtbo in the KERNEL_DEVICETREE has become
problematic as we start testing the linux-next and 6.1 as part of our
LTS migration. To address this issue we are adding in two step method
for managing the KERNEL_DEVICETREE.
Going forward we will only accept dtb/dtbo in KERNEL_DEVICETREE if it is
available in upstream. This way we ensure that the variable is more
accurate for whichever kernel you might be looking at. We have also
added a new variable KERNEL_DEVICETREE_PREFIX which our kernel recipes
will use to auto set KERNEL_DEVICETREE based on what files are in the
kernel and not a fixed list in the conf files.
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Signed-off-by: Andrew Davis <afd@ti.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
devtool has a long-standing bug where it cannot handle references to ${WORKDIR}
due to externalsrc:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=12677
linux-ti-mainline clones additional ti-upstream-tools repo alongside the kernel
sources into ${WORKDIR}, which exposes this devtool bug.
Work around the issue by moving ti-upstream-tools inside the kernel source tree,
i.e. inside ${S}
Reported-by: Radoslav Pesek <radoslav.pesek@microstep-mis.com>
Signed-off-by: Denys Dmytriyenko <denys@konsulko.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>
Extra sub-layers will be created next and relevant content moved across them.
Signed-off-by: Denys Dmytriyenko <denys@konsulko.com>
Signed-off-by: Ryan Eatmon <reatmon@ti.com>
Andrew Davis
Ryan Eatmon
Denys Dmytriyenko
LCPD Automation Script