Age | Commit message (Collapse) | Author |
|
It's common for external kernel modules to publish UAPI headers which
need to be sanitized prior to delivery to userspace. Currently, the
DDK has no way to package UAPI headers this way.
Add a new rule, ddk_uapi_headers(), which allows external modules
to package their UAPI headers for userspace. For example:
ddk_uapi_headers(
name = "my_headers",
srcs = glob(["include/uapi/**/*.h"]),
out = "my_headers.tar.gz",
kernel_build = "//common:kernel_aarch64",
)
Bug: 313898637
Bug: 326633656
Change-Id: I1b7b59def1bf0b99b1073600e1324c21f82996e8
Signed-off-by: John Moon <quic_johmoo@quicinc.com>
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
[isaacmanjarres: Used config_env_and_outputs_info and
get_setup_script, since KernelSerializedEnvInfo is not
available on this branch. Also added module_scripts
and part of the inputs for running the rule.]
|
|
Add a full checkpatch run that ignores nothing but doesn't fail on any
violations. This will be used by an AyeAye analyzer that analyzes
checkpatch violations.
Bug: 321067108
Test: Verified checkpatch_full.log file is generated with all
violations.
Change-Id: Iaad7a6368807c078b8f6461e807af1e9c538657e
Signed-off-by: Erik Staats <estaats@google.com>
(cherry picked from commit f25e1ffb33962382b15d468c19d5651eb2df40d2)
|
|
Replace the usage of the os.path.* methods with Path()
objects in the initramfs_modules_lists_test.py file to
conform to more modern python practices. Also, include
more style fixes.
Bug: 323710246
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from https://android-review.googlesource.com/q/commit:d7a39fe1a99e6c78cb3e5e57c9e742f46f35a611)
Merged-In: Ib92fc6bc3a41186f21925a45f8d53e0b5f349803
Change-Id: Ib92fc6bc3a41186f21925a45f8d53e0b5f349803
|
|
Depmod generates the modules.dep file which contains the dependencies
for each module in the directory that depmod runs in. When specifying
module lists, it is important that there's an entry for each module
in each module list to ensure that the module's dependencies are loaded
before attempting to load the module.
Therefore, add a test to ensure that each module in each module list
has an entry in modules.dep.
Bug: 323710246
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from https://android-review.googlesource.com/q/commit:2cef65c72ee9d91ebbbe184a47b1186a0954a950)
Merged-In: Ie6a7cb65cad336d4ccda9fa6957624448690e3c7
Change-Id: Ie6a7cb65cad336d4ccda9fa6957624448690e3c7
|
|
Add a test to ensure that only the expected *modules.load* lists are
generated and no extraneous lists are generated.
Bug: 324304750
Bug: 324356826
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from https://android-review.googlesource.com/q/commit:580b1a61b6a4a4f50c25f3cf3349f49099b1d9cc)
Merged-In: I7aab46022a3fcb5f754d6e517a194d0f553f21e0
Change-Id: I7aab46022a3fcb5f754d6e517a194d0f553f21e0
|
|
The initramfs.img can be either gzip or lz4 compressed, so ensure both
compression formats can be handled.
Bug: 324304750
Bug: 324356826
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from https://android-review.googlesource.com/q/commit:c4a8127d9d6606963e4605aed8764dc6c6ca2751)
Merged-In: I18655c85a329ea7a169154213739e7b9850aacd8
Change-Id: I18655c85a329ea7a169154213739e7b9850aacd8
|
|
The initramfs modules lists test is currently not comparing
the lists for equality; it is merely doing a check for None == None,
which is vacuously true. This is because the assertion compares
the return values of the sort() method on two lists.
The sort() method does not return anything, and instead sorts the
lists in place, so use the assertCountEqual() method instead, so
that the order of the lists is not a concern.
Also, fix whitespace issues and coding style when reading the lists.
Fixes: Fixes: d08605f4c795 ("kleaf: Add support for specifying modules for recovery/charger")
Bug: 266752750
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from https://android-review.googlesource.com/q/commit:3c47d6386f4a502272913d00e18b8b62c9fba9ef)
Merged-In: I6582bcc04efa5542725ca9be4df420b12919ebe0
Change-Id: I6582bcc04efa5542725ca9be4df420b12919ebe0
|
|
Commit d08605f4c795 ("kleaf: Add support for specifying modules for
recovery/charger") accidentally made it so that
modules.load.[recovery|charger] are only generated if the vendor_boot
or vendor_kernel_boot images are made. Similar to modules.load, it is
possible to want to generate modules.load.[recovery|charger] without
a vendor_boot or vendor_kernel_boot image, so decouple the two.
Bug: 324356826
Fixes: d08605f4c795 ("kleaf: Add support for specifying modules for recovery/charger")
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from https://android-review.googlesource.com/q/commit:7de44ada9d71869e582c42d5d2f62c714d496a5c)
Merged-In: Idf9db5954e8c1bae01915f104c704302dbfd81ae
Change-Id: Idf9db5954e8c1bae01915f104c704302dbfd81ae
|
|
Commit d08605f4c795 ("kleaf: Add support for specifying modules for
recovery/charger") accidentally made it so that kleaf always tries
to generate the vendor[_kernel]_boot.modules.* lists, even when
neither vendor_boot.img or vendor_kernel_boot.img need to be built.
Fix this by only attempting to generate those artifacts if either
vendor_boot.img or vendor_kernel_boot.img need to be built.
Bug: 324304750
Fixes: d08605f4c795 ("kleaf: Add support for specifying modules for recovery/charger")
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from https://android-review.googlesource.com/q/commit:100076d823899780662a6fdec71f92be35f6d9f7)
Merged-In: I31ec16fc543e55f33319e017c7d5976e825f581d
Change-Id: I31ec16fc543e55f33319e017c7d5976e825f581d
|
|
* Additionally include ddk_module as suggestion for build external
kernel modules.
Bug: 306122000
Change-Id: I15eaefc4a205c23c22b3db26d9d1e31654507e56
Signed-off-by: Ulises Mendez Martinez <umendez@google.com>
(cherry picked from commit d6e4d2543f2364be1283e096bda49516da10a68b)
|
|
Bug: 306122000
Change-Id: I43d17a45caf81e2be7a281bd8498baa0b7b2ef80
Signed-off-by: Ulises Mendez Martinez <umendez@google.com>
(cherry picked from commit cf40b51691f3690a198c6aaeccb55baa8d4cc0b0)
|
|
* This is done by keeping a local copy of the make generated one, for
each ddk_module or kernel_module. For kernel_module_group or
kernel_modules_install this becomes a depset which is traversed in
postorder (dependencies first) when needed.
* With this change kernel_images (initiramfs, (system|vendor)_dlkm) can
control whether to use this new approach or keep the old behavior with
the variable `create_modules_order` (set to True by default).
Bug: 306122000
Change-Id: Ic636c2da8d77e6c25e7eb9c8f6de68904828a9ad
Signed-off-by: Ulises Mendez Martinez <umendez@google.com>
(cherry picked from commit 1d3379dd0e0bb660bff2903ce7e93bd93e696700)
|
|
The system_dlkm partition consists of GKI modules, which are not loaded
during first-stage init. However, the MODULES_[RECOVERY|CHARGER]_LIST
arguments specify vendor modules that must be loaded during first
stage init. Therefore, it does not make sense to consider these
lists when building system_dlkm.img, so do not consider them.
Bug: 322408856
Bug: 323710246
Change-Id: I50fb42b40a94b0b53d5a2cda676d39b919395538
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from commit 4daab96c0623f571c0eefbb0a4fe2d955e86fdea)
|
|
When handling charger and recovery modules lists, depmod
is run on the set of charger and recovery modules and all
modules. However, the order in which this happens is such that
depmod is run on the set of charger modules, instead of all modules
last. This means that the final modules.dep file will contain
the dependencies for the charger modules, and not the rest of the
modules.
This is not ideal, as the modules.dep file should contain the
dependencies for all modules. Thus, ensure that depmod is run
on all modules last after running depmod on the recovery and
charger modules lists.
Bug: 322408856
Bug: 323710246
Change-Id: I7e03be016877df3b528b810735efec6c55fd0f6e
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from commit 3f67b5ffe1348d438e47f95fe2f2250def080960)
|
|
The variables pertaining to the recovery and charger lists were
not written correctly, so fix them.
Bug: 322408856
Bug: 323710246
Change-Id: Ia4d18920c5460a6ab019cb186e9da5cb1dd8835a
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from commit 8565b2e637df5d0a482426b9aa505c917b9711aa)
|
|
As the common_kernel_sources filegroup includes _any_ file in the tree,
with the exception of Git and Bazel files, the build system gets
confused if some files are present at the time where the glob is
evaluated but absent in following build commands. An example of this is
an IDE automatically updating the code index file when saving a source
file (e.g. tags.lock or tags.temp) as the kernel is getting built.
Given that cscope and tags files are officially supported by the Linux
Makefiles and as Bazel doesn't actually need them to build the kernel,
exclude them from the filegroup.
Test: build the kernel
(cherry picked from https://android-review.googlesource.com/q/commit:7a3b2fbab0d764d6f6acbcf36c7b45f02421cf55)
Merged-In: Id103c3f22dc286960fd703aac2c8e527cb718ff5
Change-Id: Id103c3f22dc286960fd703aac2c8e527cb718ff5
|
|
When building a ddk_module with --config=local:
- OUT_DIR is in out/cache/<hash>/$KERNEL_DIR
- Module.symvers of the depended module is restored to
out/cache/<hash>/<package>/<name>_Module.symvers
- However, the generated makefile expects the file
in COMMON_OUT_DIR/<package>/<name>_Module.symvers,
where COMMON_OUT_DIR is not changed.
This is because cache_dir.bzl only modifies the
value of OUT_DIR, not COMMON_OUT_DIR.
Hence, it makes sense to set COMMON_OUT_DIR to the
directory under cache_dir as well.
Test: TH
Bug: 304499217
Change-Id: Icd9beb73db7105b278def0c95f036cd1ae21a930
Signed-off-by: Yifan Hong <elsk@google.com>
(cherry picked from commit e2a45faccec7c5d94b4ff9d2c2d11d5d8d286814)
|
|
When ddk_module is building with --config=local:
- OUT_DIR is in out/cache/<hash>/$KERNEL_DIR
- Module.symvers of the depended module is restored to
out/cache/<hash>/<package>/<name>_Module.symvers
- However, the generated makefile expects the file
in COMMON_OUT_DIR/<package>/<name>_Module.symvers,
where COMMON_OUT_DIR is not changed.
This is because cache_dir.bzl only modifies the
value of OUT_DIR, not COMMON_OUT_DIR.
Hence, let kernel_module's command to restore its
Module.symvers to use COMMON_OUT_DIR directly.
Bug: 304499217
Change-Id: I28995f30812e7ac2d596982d70e79f9374364787
Signed-off-by: Ulises Mendez Martinez <umendez@google.com>
Signed-off-by: Yifan Hong <elsk@google.com>
(cherry picked from commit 2f12101856d947353d043aec76ed0bf7c09e1c70)
|
|
Revert submission 2922866
Reason for revert: Retargeting to `android15-6.6`, no reason to have any divergence on this branch.
Reverted changes: /q/submissionid:2922866
Change-Id: If584557f5e971ec232b4eecf58d901015c954138
|
|
Revert submission 2922866
Reason for revert: Retargeting to `android15-6.6`, no reason to have any divergence on this branch.
Reverted changes: /q/submissionid:2922866
Change-Id: I43c56b9d14cfee2d06c6284b71db08f18a971806
|
|
Bug: 284390729
Change-Id: I1d2691500548beb2f50b3ce4250eef78ebc35f71
Merged-In: I1d2691500548beb2f50b3ce4250eef78ebc35f71
|
|
Now that Roboleaf will stop using external.bzl, there's
no benefit to keep a common definition. Hence, just
expand the macro in workspace.bzl directly.
Bug: 293260835
Change-Id: If39846a6003d937471b70bb466621755d1165cbc
Merged-In: If39846a6003d937471b70bb466621755d1165cbc
|
|
If you include a Reported-by: tag, checkpatch will
warn with BAD_REPORTED_BY_LINK if there is not a URL link
to where the report was made.
While these links are nice, and should be included where
possible, they shouldn't be required as often reports come
in private email, and we should make sure to credit reporters
even if there isn't a public URL for that report.
Change-Id: I1a06a5c28c3dd9f7b365b475cbb446ba6d48c13c
Signed-off-by: John Stultz <jstultz@google.com>
|
|
microdroid-gki is maintained separately. Ultimately we may want to move
microdroid-gki to somewhere else. Until then we should leave it as-is
when updating normal GKI prebuilts.
Bug: 317929593
Test: run script
Change-Id: I1fc3f2c7c2e68f9121f2c7d38cb7a4372528557d
(cherry picked from commit 4e49573f0aa24198acaadab789595377f799e90e)
|
|
Fix errors like:
output 'build/kernel/kleaf/tests/ddk_test/makefiles_test_suite_subdir_test_subdir_sources_same_name_subdir_kbuild/subdir/Kbuild' is a dangling symbolic link
Test: bazel test //build/kernel/kleaf/tests
Bug: 319893821
Bug: 319077326
Change-Id: I445c3e49f56cdf82bbae7760b8abbe77b985a189
(cherry picked from commit ba157874458502fad044dd87342d1634a4cb1bf7)
|
|
As of bazel's e2c0276d1fd6 ("Enable
`--incompatible_sandbox_hermetic_tmp` by default"), sandboxes can be
much more neat, e.g. /tmp/bazel-working-directory/_main. Add support for
that new behaviour when checking whether we are inside a sandbox.
Link: https://github.com/bazelbuild/bazel/commit/e2c0276d1fd69c19d6946fb17f9e983e9ea68025
Bug: 319077326
Change-Id: I51498306fa620626d217dd89661c7acaba25560b
Signed-off-by: Matthias Maennich <maennich@google.com>
(cherry picked from commit 98b5f4865e13e39e5f6d973cf1f71752396073c3)
|
|
Previously without the flag, the TreeArtifact contains symlinks to
files under execroot, which is allowed.
With the flag, the symlink points to a file in the sandbox, which
is no longer valid (the sandbox is teared down) when the TreeArtifact
is validated.
Change it to a cp so that it is straightforward to humans.
Performance: There are no obvious performance regression associated with
this change. The KbuildMixedTree action is on the critical path.
This change reduces the time for the action from 4.1s to 3.6s in a
one-time experiment.
Test: TH
Bug: 319893821
Bug: 319077326
Change-Id: I63f2364e0ffaa57a7ae57a1f7730f4b510ae3a8e
Signed-off-by: Yifan Hong <elsk@google.com>
(cherry picked from commit b626eb43be646d8efdfb57e7d832da39fd81284a)
|
|
Fixes: a09156147f79 ("kleaf: allow build number in downloaded file name")
Bug: 317224407
Change-Id: I222c49ebcd7b582d1f33a1afd55f165ea1679b58
Signed-off-by: Ulises Mendez Martinez <umendez@google.com>
|
|
* This is needed to allow generated files, which otherwise are not
discoverable as inputs.
Bug: 319236725
Change-Id: Iaf7a3f85f3d5d0bde4a4ddce140f94cb873b860c
Signed-off-by: Ulises Mendez Martinez <umendez@google.com>
|
|
Kbuild only accepts .S not .s. Bazel suffix
matching is case-sensitive.
Test: build trusty
Change-Id: I645c226f01bcd577f809c1759b9e71541d876522
|
|
mkbootimg.py supports specifying the kernel base, as well as other
offsets that a bootloader can read from the vendor boot header.
Therefore, add support for specifying these offsets when building
the vendor_boot image.
Bug: 308443544
Change-Id: I9da3284a93ee064d07034d3a7a3e42c07e0e9b88
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
|
|
Currently, there are two places where "pathlib.path" is used instead
of the correct "pathlib.Path". This causes the following error when
building with --ignore_missing_projects:
AttributeError: module 'pathlib' has no attribute 'path'. Did you mean: 'Path'?
Fix this by changing "pathlib.path" to "pathlib.Path".
Bug: 306026132
Change-Id: I8ecdec3b48d7438a3a953944f629b462a7c88536
Signed-off-by: John Moon <quic_johmoo@quicinc.com>
(cherry picked from commit 224cf84e94caf3db491139ce2dac462d97a833e3)
|
|
Make it clear that setting TRIM_UNUSED_MODULES removes unused
module binaries from the final build output.
Bug: 311553435
Change-Id: Iad5eac931381d76a2f9fe2233b9305bc55775a63
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
|
|
The current regex for detecting comments in a module list escapes the
"#" character. The "#" character is not a special character so it does
not need to be escaped.
Bug: 311553435
Change-Id: If295a0e14b922ebefc4e09cafad252379977a8f8
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
|
|
Currently, the same modules.load file is used for normal boot,
recovery/fastbootd and charger modes during first-stage init. This
means that during normal boot, the modules that are needed for
recovery and charger modes are loaded during first-stage init, which
is not required.
Add support for specifying a list of modules that are required for
booting into recovery or charger mode. This introduces two new
attributes to the kernel_images() macro called: modules_recovery_list
and modules_charger_list. Each attribute points to a list of modules
that are needed to boot into each mode respectively. This list can
be generated dynamically by concatenating the list of vendor_boot or
vendor_kernel_boot modules required for first stage init, along with
the list of modules needed to boot into each mode.
Defining these attributes also produces a file named
modules.load.[recovery/charger] in the initramfs, which init uses
instead of modules.load when booting into recovery or charger mode
respectively to load the first-stage init and recovery and charger
modules. In case of normal boot, init will use the modules.load
file, which will only contain the modules needed for first
stage init, and the recovery and charger modules will be loaded
from the vendor_dlkm partition during second-stage init.
Bug: 266752750
Change-Id: I11503b11683ef64cf0933b8641959ea5acd63ab1
Signed-off-by: Isaac J. Manjarres <isaacmanjarres@google.com>
(cherry picked from commit b9ba4b67e90b85d01eb2f9cebf62830a73574cd9)
[isaacmanjarres: Re-wrote some of the commit message/implemented it
as part of kleaf. Also added tests for the new attributes.]
(cherry picked from commit d08605f4c795e05d8569e975affd2736e430a555)
[isaacmanjarres: Resolved trivial merge conflicts when backporting
from main.]
|
|
Drop references to
//build/kernel:hermetic-tools/...
Such aliases have been deprecated. This ensures
that running these tests won't produce warning
messages because they depend on the deprecated
aliases.
To achieve this, I defined py_binary for each
test script, then replaced py_test with hermetic_exec_test
that invokes the script with the given arguments.
in PATH before running the test. Then I modified
the tests to search for the binary in PATH.
Test: TH
Test: bazel test //common-modules/virtual-device:x86_64/goldfish_drivers/goldfish_sync_test //common:kernel_aarch64_test //common:kernel_aarch64_additional_tests --test_output=all --config=stamp
Bug: 234125794
Change-Id: Id84321227e6af04e2433516c30f8b11cbb9bcafe
|
|
The SPL in the platform build is being updated more frequently than
originally anticipated (previously it was updated quarterly in the
trunk_staging builds). Now the trunk staging SPL follows the monthly
SPL. To accomidate this, this patch updates the development build's SPL
to pick the QPR date after the next QPR, e.g. Nov 2023 builds would pick
March 2024.
Note, as a reminder, this is only applied for development GKI boot
images (not certified GKI boot images).
Bug: 274825778
Change-Id: I4b746e417d8b0403ad7a2cb21a83001cd1ecbd40
Signed-off-by: Will McVicker <willmcvicker@google.com>
|
|
The flag fetches kernel_<arch> and updates the prebuilts at
kernel/prebuilts/<version>/<arch>/microdroid-gki.
Bug: 304479739
Test: build/kernel/gki/download_from_ci -b 304479739 \
--update-microdroid-gki 11062399
Change-Id: Ibd1d28d8d61c4d2160483dec2c35bb2b16072a21
|
|
If a line is trailing with # nocheck, e.g.:
CONFIG_FOO=y # nocheck: b/???????
Then this config will not be enforced in the
final .config. A warning is printed if value
does not match.
Bug: 295240936
Bug: 286589887
Change-Id: I12b3c16e14938116cc80af5abead08e35433eb2d
(cherry picked from commit 99119c8fdeddba644c3d741b88341ca26f9d662d)
|
|
Enabling building both erofs and ext4 image changed the
legacy image from system_dlkm.img to system_dlkm.ext4.img.
This may break the scripts downloading this image from
kernel build artifacts. For backward compatibility on the
released kernel artifacts name ext4 images with legacy
name system_dlkm.img.
Fixes: ebfbdd4073c7 ("kleaf: system_dlkm: Build ext4 & erofs images for common kernels")
Bug: 297407875
Test: TH, Manual image verification in dist
Change-Id: Ia5aa7be37eff2801dfd50e8354213c2614465242
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
|
|
For common kernel targets build system_dlkm partition
image for both supported read-only filesystems to
make them available via CI build artifacts by default.
Before this change:
ext4: system_dlkm.img
After this change:
ext4: system_dlkm.ext4.img
erofs: system_dlkm.erofs.img
Bug: 297407875
Test: TH, Manual image verification in dist
Change-Id: Ibad393af4fcd7ba10f6dec49a3110ea5c163902f
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
|
|
The following is implemented:
* Add hashtree footer to vendor_dlkm.
* Add SPL to boot.img
This allows developers to flash a custom kernel on an unlocked device
without having to wipe the device (based on SPL) or disable verification
(based on AVB hashtree requirement).
Test: verify custom boot.img and system_dlkm.img boots without wiping on C10
Test: verify custom vendor_dlkm.img boots without wiping on C10
Bug: 274825778
Change-Id: I685199eb11bfc5a4bdf545170ec136b327f992ad
Signed-off-by: Will McVicker <willmcvicker@google.com>
|
|
Add /system/lib/modules/ prefix for every modules
in modules.dep file for flatten images. Flatten
images have directory structure as /lib/modules/*.ko;
and when system_dlkm is mounted modules are available
to load from /system/lib/modules.
Test: Manual modules.dep verification, TH
Bug: 301178664
Change-Id: Ife5dad13b3a51632d1fe9000474c4fff062e4fc3
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
|
|
Enable generating flatten system_dlkm images for the
common_kernels targets to make them available for CI
builds for testing.
Bug: 301178664
Test: Manual image verification with different options
Test: TH
Change-Id: I08745d7660ab232ad92156ec764cc20ef749adad
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
|
|
Load statement is out of its lexicographical order
error fixes.
Test: TH
Change-Id: Id775d2d6cd6ec77044490f16b954dbf1dbecf098
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
|
|
Generate flatten images if kernel_images have
build_system_dlkm_flatten is set True. These
images don't have the `uname -r` in the directory
structure and all modules are available as
lib/modules/*.ko with modules.load and modules.dep
depmod artifacts.
Generates images as follow:
If system_dlkm_fs_type is set:
system_dlkm.flatten.<system_dlkm_fs_type>.img
If system_dlkm_fs_types is set:
system_dlkm.flatten.<fs_type>.img for each fs types
By default when no fs type is specified:
system_dlkm.flatten.ext4.img
Bug: 301178664
Test: Manual image verification with different options
Test: TH
Change-Id: I4b7815d131d3d26adaa35c4550e76db71aa33936
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
|
|
Introduces system_dlkm_fs_types list attribute for the
kernel_images rule to build the system_dlkm partition
images for all filesystem provided in the list.
Deprecates the system_dlkm_fs_type but gives it priority
for backward compitability resons.
Behavior based on values of system_dlkm_fs_type &
system_dlkm_fs_types:
Both specified: Fail with deprecation message
Both not specified: system_dlkm.img with ext4
This is same as current default behavior when
system_dlkm_fs_type is not specified.
system_dlkm_fs_type="fs": system_dlkm.img with given fs
This is same as current behavior.
system_dlkm_fs_types=[list_of_filesystems]: system_dlkm.fs.img
for each file system type in list_of_filesystems
Bug: 297407875
Test: TH, Manual with all permutations & combinations
Change-Id: Ie8ee880dbbc41576603a1ab1b56f0307a6ab2e90
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
(cherry picked from commit b0fedfdcff7d4e8fcb6ec82ac8294e738b78dcbc)
|
|
--update-16k fails fetching the version and SHA
information as version parsing logic only supports
format for x86_64 whereas 16K page size kernels
are only available for the arm64 architecture.
Since kernels for all architecture targets are built
from the same source at particular git hash; use
x86_64 target's kernel binary from the same CI build
to fetch the kernel version and SHA for --update-16k.
Bug: 296248256
Test: download_from_ci --update-16k -b 296248256 10809747
Change-Id: I9514da8392703a7478ea3dd2c1425ef0ca31ab9f
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
|
|
The build_errors.log has always been too long,
making the infra to skip lines in between, which
usually contain useful error messages.
Bug: 296431087
Change-Id: I2f71af242ba64e3dff923ef5f469f0faa0068408
(cherry picked from commit 4025481becbd2f07ec9bebd7556504c86379dd25)
|
|
Fixes buildifier error:
Load statement is out of its lexicographical order.
Change-Id: I66b6ee21802c012a3834b71e37c238b18296e6a2
Signed-off-by: Ramji Jiyani <ramjiyani@google.com>
(cherry picked from commit 89ee0ed2f6a536a6762a778e0d5910a442fb425b)
|