ACPM is needed for the S2MPG10 PMIC, and simple-fb for some basic
display.

The expectation is that ACPM related kconfig symbols will be enabled in
the future by the kernel defconfig directly, but simple-fb should never
be enabled, as it's deprecated. We can enable it here until we have a
working display driver.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

* add maxim max77759 gpio, nvmem, fuel gauge
* reboot mode in nvmem
* do cold-reset

Signed-off-by: André Draszik <andre.draszik@linaro.org>

This is next-20250327 with ACPM, S2MPG10 (RTC), and UFS stability fixes
on top.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

We need this to mark boot slots as successful, etc. and at least for
now it seems useful to have it in initramfs for testing and playing
around.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

We need this to mark boot slots as successful, etc.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

This is a quick hacked together version of Pixel's bootctl for Android
without requiring Android or any AIDL for setting boot slot and boot
success etc.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Signed-off-by: André Draszik <andre.draszik@linaro.org>

The otherwise weak default linux-linaro-googlelt is a bit outdated at
this stage (it's at v6.10-rc4).

Signed-off-by: André Draszik <andre.draszik@linaro.org>

This reverts commit a26eb8a9.

yocto v5.1 (styhead) has the v6.10 kernel, which does have support for
Google GS as that exists since v6.8.

This means we can enable the linux-yocto recipe for Google GS.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

We don't want to power off immediately upon a power button press, as
it's too easy to do that accidentally.

Update to require a 5 second button press.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

This fixes a build error due to Yocto having removed support for lz4c
(which has been deprecated since at least 2018).

Signed-off-by: André Draszik <andre.draszik@linaro.org>

This will apply any necessary fixups for which we don't have (upstream)
drivers yet.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

This is a gross hack, but allows us to tweak some registers without
having drivers.

This initial use is to enable the hardware trigger on decon, to have
framebuffer updates visible on the display.

It relies on busybox' devmem utility, which we've enabled in
commit eddfb28f ("busybox: busybox_extra.cfg: add watchdog and
devmem utils").

Signed-off-by: André Draszik <andre.draszik@linaro.org>

In some environments, the default umask doesn't allow bitbake to run,
and it complains:
    Please use a umask which allows a+rx and u+rwx

Let's do that, to enhance the out-of-the-box experience in such cases.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

We need to follow upstream poky commit 263468c351c6 ("bitbake.conf:
make ZSTD_COMPRESSION_LEVEL a number rather than a dash-prefixed
option") and update our use of ZSTD_COMPRESSION_LEVEL.

Upstream uses it now like such:
    ZSTD_COMPRESSION_LEVEL ?= "3"
    ZSTD_DEFAULTS ?= "--threads=${ZSTD_THREADS} -${ZSTD_COMPRESSION_LEVEL}"
so we need to update the assignments accordingly.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Upstream poky commit 43b8b3fa72d7 ("classes-recipe/core-image: drop
debug-tweaks IMAGE_FEATURE") ") has removed debug-tweaks as an image
feature, and instead implied features need to be enabled explicitly
now.

Do so and replace it with:
    allow-empty-password empty-root-password allow-root-login

Note that post-install-logging was also implicitly selected, but I
don't tink this is useful by default, so we leave it out.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Upstream poky commit 43b8b3fa72d7 ("classes-recipe/core-image: drop
debug-tweaks IMAGE_FEATURE") has removed debug-tweaks as an image
feature, and instead implied features need to be enabled explicitly
now.

Whitelist sysrq if both, allow-root-login and empty-root-password, are
set in IMAGE_FEATURES, as it seems that if root is allowed to log-in
without password, we might as well enable the sysrq.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Upstream poky commit 54dd5c56fc76 ("classes: rootfs-postcommands:
autologin root on serial-getty") (part of the langdale release) added a
similar feature as we have here.

Switch to testing for it in IMAGE_FEATURE_WARNINGS and drop the custom
ROOTFS_POSTPROCESS_COMMAND that was used for it.

Note1: Our version didn't work for SysVinit as of upstream poky
commit 2e608fd1b70f ("sysvinit-inittab: support non-busybox-getty on
serial consoles")

Note2: The upstream version doesn't correctly deal with
SysVinit+getty(busybox) I believe, but we don't care at this point in
time, sysvinit support in our version was broken anyway.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

serial-autologin-root is the upstream replacement for our custom
autologin image feature, which is about to be removed.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Upstream poky commit 54dd5c56fc76 ("classes: rootfs-postcommands:
autologin root on serial-getty") (part of the langdale release) added a
similar feature as we have here.

Test for it in addition to our existing custom autologin feature in
IMAGE_FEATURE_WARNINGS.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

We want this enabled by default, also we can not react to USB-C cable
orientation properly.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

We want this enabled by default, also we can not react to USB-C cable
orientation properly.
While at it, reduce the delay from default 5s down to 2s.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

There is no need for this trailing space.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Sometimes kernel modules are not being copied from the initramfs to the
real root fs.

The reason is that mkbootimg (which is used to generate the initramfs)
unconditionally sets time stamps to 0 (for reproducible builds). rsync
by default uses file size and file modification time to determine if
files need to be copied. Therefore, rsync will not copy files unless
file size has changed as well.

File size not changing happens particularly often when making small
changes or e.g. fixing a typo in a string.

Therefore, instruct rsync to not only rely on the above two indicators,
but to also compare file contents by using its --checksum command line
argument. Note that rsync will only compare contents as an additional
step after it has compared time and size (but in our case time always
matches between sender and receiver, as per the above).

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Yocto has promoted these to buildpath issues to errors in
commit e665be2ad94b ("insane: Promote long standing warnings to
errors") which now fails the build:

    ERROR: linux-linaro-googlelt-6.10-r0 do_package_qa: QA Issue: File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/hdmi.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/a4xx.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/a5xx.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/dsi_phy_20nm.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/a6xx.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/adreno_common.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/a2xx.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/dsi_phy_28nm_8960.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/sfpb.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/a3xx.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/dsi_phy_10nm.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/dsi.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/dsi_phy_7nm.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/adreno_pm4.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/mdp5.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/mdp_common.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/mdp4.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/dsi_phy_14nm.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/gpu/drm/msm/generated/dsi_phy_28nm.xml.h in package linux-linaro-googlelt-src contains reference to TMPDIR
    File /usr/src/debug/linux-linaro-googlelt/6.10/drivers/tty/vt/consolemap_deftbl.c in package linux-linaro-googlelt-src contains reference to TMPDIR [buildpaths]
    ERROR: linux-linaro-googlelt-6.10-r0 do_package_qa: Fatal QA errors were found, failing task.

While we wait for upstream to fix these reproducibility issues, demote
them back to warnings temporarily(!).

Signed-off-by: André Draszik <andre.draszik@linaro.org>

The steps to add a new repository (project) are:
    1) update yocto-master.xml with the new <project name=...>
    2) repo sync
    3) build & run validation
    4) run this script

This script will update default.xml with the SHA1 of the project that
was added in step 1).

So far, subsequent invocations of this script would fail, as the
project exists twice now: in default.xml and in yocto-master.xml.

So let's also update yocto-master.xml at the same time to add a
<remove-project ...> line for the project from step 1) (which had to be
done manually previously), avoiding the extra manual intervention.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Our manifest contains four spaces at the start of the line, adjust the
script here accordingly.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Merge branches 'b4/gs101-non-essential-clocks-2', 'b4/usb-phy-gs101',
'b4/gs101-usb-regulators-in-dt', 'b4/pete-of_syscon_register_regmap'
and 'fixes' into HEAD

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Otherwise fastboot complains, at least in some environments:

    $ fastboot flash userdata core-image-full-cmdline-google-gs.rootfs.ext4
    error: write_sparse_skip_chunk: don't care size 82365440 is not a multiple of the block size 4096
    Sending sparse 'userdata' 1/2 (254972 KB)          error: write_sparse_skip_chunk: don't care size 82365440 is not a multiple of the block size 4096
    error: write_sparse_skip_chunk: don't care size 82365440 is not a multiple of the block size 4096
    OKAY [ 10.462s]
    Writing 'userdata'                                 (bootloader) chunk #21 header corrupted
    FAILED (remote: 'failed to flash partition (userdata): -7')
    fastboot: error: Command failed

Align it as required.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

* using your own kernel
* avoiding rm_work
* install extra packages

Signed-off-by: André Draszik <andre.draszik@linaro.org>

This reverts commit de7b893e.

meta-openembedded has been fixed. No need for this patch anymore.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

We don't support or require the custom initramfs-as-rootfs anymore, so
drop the custom conf-notes.txt and simply copy and use the one provided
by poky's meta-poky

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Instead of errors out only if no layers could be found, be a bit more
clever and error out if not all requested layers could be found, e.g.
because of typos.

Such behaviour is less surprising that just silently ignoring the request
for additional layers.

This change wasn't tested much, but hopefully should be ok.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

It is useful to be able to skip layer selection altogether even if
no extra layers are given, i.e. if EXTRA_LAYERS= is given. This is
different to not specifying EXTRA_LAYERS at all.

So far, the extra layer selection still pops up even if no extra layers
are desired (denoted by EXTRA_LAYERS=).

Update the code to handle that case.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

* we now depend on yocto/master at least with commit 812dafbec191
  ("base/bitbake.conf: Introduce UNPACKDIR") (due to using UNPACKDIR)

Signed-off-by: André Draszik <andre.draszik@linaro.org>

As of poky.git commit [1] ("base: Switch UNPACKDIR to a subdir of
WORKDIR"), additional sources end up in UNPACKDIR, not WORKDIR, so
update the recipe for that. Fixes the following build error:

    | Could not find kernel config fragment .../tmp/work/google_gs-poky-linux/linux-linaro-googlelt/6.9/filesystems.config
    | WARNING: .../tmp/work/google_gs-poky-linux/linux-linaro-googlelt/6.9/temp/run.do_configure.2426018:184 exit 1 from 'exit 1'
    | WARNING: Backtrace (BB generated script):
    |       #1: do_configure, .../tmp/work/google_gs-poky-linux/linux-linaro-googlelt/6.9/temp/run.do_configure.2426018, line 184
    |       #2: main, .../tmp/work/google_gs-poky-linux/linux-linaro-googlelt/6.9/temp/run.do_configure.2426018, line 306
    ERROR: Task (.../meta-bsp-google-gs/recipes-kernel/linux/linux-linaro-googlelt_6.9.bb:do_configure) failed with exit code '1'

Link: https://git.yoctoproject.org/poky/commit/?id=d24a7d0fb16457e10e7a216d4c9ae3fb265113d1

 [1]
Signed-off-by: André Draszik <andre.draszik@linaro.org>

Fixes the following build error:
    ERROR: .../meta-bsp-google-gs/recipes-core/initrdscripts/initramfs-framework-rootfssize.bb: Using S = ${WORKDIR} is no longer supported
    ERROR: .../meta-bsp-google-gs/recipes-core/initrdscripts/initramfs-framework-modulescopy.bb: Using S = ${WORKDIR} is no longer supported
    ERROR: .../meta-bsp-google-gs/recipes-devtools/android/android-tools-adbd-cmdline.bb: Using S = ${WORKDIR} is no longer supported

Do the same as poky.git in commit [1] ("recipes: Switch away from
S = WORKDIR").

Link: https://git.yoctoproject.org/poky/commit/?id=d6ae8d2004f8ac3aab4571e0077d406b2d19b542

 [1]
Signed-off-by: André Draszik <andre.draszik@linaro.org>

Upcoming changes (UNPACKDIR) mean that this will become incompatible
with scarthgap.

Signed-off-by: André Draszik <andre.draszik@linaro.org>

Signed-off-by: André Draszik <andre.draszik@linaro.org>

* meta-openembedded/meta-oe is not optional anymore due to
  template/local.conf pulling in adbd-related packages
* building just the kernel alone is not enough anymore since we expect
  to mount a normal rootfs from flash now

Signed-off-by: André Draszik <andre.draszik@linaro.org>