Currently, ota_from_target_files(a python script) calls
brillo_update_payload(a bash script), which then calls delta_generator(a
C++ binary) to do the actual diffing. Having brillo_update_payload in
the middle does not offer any additional flexibility, but it makes
maintaince more difficult. Bash code is less readable and harder to
debug. Further more, everytime we want to add a new flag to
delta_generator, we have to add the flag in three places:
ota_from_target_files, brillo_update_payload, delta_generator.
Historically, brillo_update_payload was there because we inherited from
ChromeOS side. This bash scripts extracts target_files.zip and invoke
delta_generator to generate the actual OTA. Any customizations we want
on OTA must be implemented by modifying the input target_files.zip ,
Manipuating big zip files is slow and inefficient.
To make thing simpler/faster, remove the intermdiary call to
brill_update_payload. ota_from_target_files will now extract target
files and call delta_generator directly.
Test: th
Bug: 227848550
Change-Id: I44b296e43bf7921bcf48ef6a1e4021a12669739c
'bsdiff' tool should be used when recovery and boot image
have mismatching entries. This includes not only the case
when system_as_root enabled, but include_recovery_dtbo and
include_recovery_acpio too.
Add check if they are enabled to synchronize logic with
core/Makefile, which already takes this into account.
Test: make dist -j
Change-Id: Ie673403ae8eae446d006cc524dd48ac6c56194fc
Signed-off-by: Oleg Lyovin <ovlevin@sberdevices.ru>
Instead of relying on image build flags, check the actual file to
determine if input is sparse image.
Test: th
Change-Id: I6e58603535039eb3c4e799c61ed9f4783f1888dd
Previously ("Add null and console dev nodes to the Android initramfs")
introduced into init_boot.img the /dev/console and /dev/null nodes.
Sadly, it missed changing the release tools which are also generating
images in their own way.
Bug: 254835242
Change-Id: I4d2893ac24c09cf0d117422cbe2bde66083e7118
ZipDelete() works by copying every non-deleted entry to a new zipfile.
Current implementation uses python's zipfile module to perform entry
copying, which is inefficient, as every entry must be decompressed and
then re-compressed. Instead, use zip2zip which avoid re-compression.
Improvement: deleting META/dynamic_partitions_info.txt from
raven-target_files-9465001.zip improved from 500+ seconds to 13 seconds.
Change-Id: I0548255bc29380303314763f6d81e74bf3dbb76e
Currently we only have vbmeta_system and vbmeta_vendor. Add
BOARD_AVB_VBMETA_CUSTOM_PARTITIONS , which allows OEMs to define
their own vbmeta_*.img partitions. example: aosp/2381823
Bug: 263568805
Change-Id: Id671e2c3aee9ada90256381cce432927df03169b
In the old days, we hacked values of ZIP64LIMIT to get around size
limitations of non-zip64 supported zip files. Now that we switched to
python3 + zip64, there's no point in keeping those hacks.
Test: th
Bug: 255683436
Change-Id: I913db33dad5503736c68a7a1f1321aa952019f60
When performing a deep copy of info dict, input_file is serialized as
filename. Make ExtractFromInputFile support reading from filename of a
ZipFile.
Test: build OTA with partner provided target_files
Bug: 253549364
Change-Id: I366a076c32d638d61b83a1df9bf864807cf5257c
When trying to make a deep copy of PartitionBuildProps, execution fails
with picle error because PartitionBuildProps stores a ZipFile object,
which stores a lock object, which doesn't support pickling.
To fix it, pickle ZipFile object as a str to its path.
Test: generate ota with boot variable files
Bug: 253549364
Change-Id: I92c5d78ce65045495203c85862922d33d886e9ea
'zip' is a host command, and it can not be provided as a dependency when some release tools are used in Bazel rules.
Test: atest --host releasetools_test
Bug: 243748589
Change-Id: Ie5f42eadbd316ccd018b19194c466a908971af82
Care maps need to store the original image size, which excludes bytes
used by hash tree or FEC code.
We used to propagate original image size using the global OPTIONS
dictionary. This is bad coding practice, and also fragile because we
have to make sure what's stored in dictionary and what's on disk are
consistent. Instead, let's read the content of images on disk, and parse
the AVB footer. The AVB footer contains the ground truth original image
size.
Test: build OTA, make sure the care maps have valid range
Bug: 246504616
Change-Id: I9250b478ab34dda60578a6b8c23ae6d7a9385788
Some partners have large number of products that share common
images. Generating OTAs for all these products waste a lot of CPU
cycles, because we waste time diffing the same images multiple times.
To mitigate this, add a tool for merging partial OTAs. Partners can
generate a partial OTA for common images, and re-use the generated OTA
by merging with each product specific OTA.
Bug: 227848550
Test: Run merge_otas.py on 4 OTAs, install the generated OTA
Change-Id: Ie38d522cdc9b89b1123b3d8190516c3648f99788
* changes:
Remove HashTreeInfo from verity_utils.py
Remove verity keys from info_dict
Remove replace verity key args
Remove unused args from common.py
Clean up unused code for VB in verity_utils.py
Remove verity related props from build_image.py
We are removing VB related functionalities from release tools. This
change removes the hashtreeinfo and generator in verity_utils.py. It
also clears the function args using it.
Bug: 241044073
Test: atest under build/make
Change-Id: Icc298256be6917ffbd40cf4735f8e8092b541c9f
We are removing VB support from release tools. This change aims to
remove verity keys in info_dict generated by common.LOADINFODICT
Bug: 242672222
Test: atest under build/make
Change-Id: I9268cb113c2c0bd11ce85b8485e5a4605bdbdca0
We are removing support of VB from release tools. This change aims to
remove args that no longer be used.
Bug: 242672222
Test: atest under build/make
Change-Id: I347447ebf1e589f8d3c4c8254f2940911c0be7f1
BUG:242954151
Test: make target-files-package, then check the two sha1 are same:
1. sha1 of IMAGES/boot.img
2. expected sha1 of boot in VENDOR/bin/install-recovery.sh
Test: make target-files-package; unzip target_files.zip IMAGES/\*`;
`zip -d arget_files.zip IMAGES/\*`
`add_img_to_target_files -a target_files.zip`.
Verify that the rebuilt images are identical to the deleted ones.
Change-Id: I5906ac1696b6403674fd8330b1569aed53510945
Uniform the split() function
str.split() will return a list split all spaces in str,
while str.split(' ') will return a list might contain ''
which might have potential issues.
Signed-off-by: jiajia tang <tangjiajia@xiaomi.com>
Change-Id: I0961659b140f800bdbe285f63bb4f02b8459ff8b
Signed-off-by: jiajia tang <tangjiajia@xiaomi.com>
Some OTA files may have additional OTA props; when this is being computed within
ota_utils.py -> ComputeRuntimeBuildInfos, the ramdisk type needs to be passed into
PartitionBuildProps.FromInputFile as not all ramdisk may still use gzip. This check
is already defined in common.py, so this change also exposes the function GetRamdiskFormat.
Bug: 231075507
Signed-off-by: TJ Rhoades <tjr@microsoft.com>
Change-Id: If1b93b887990f0d90df2c6003122821c1e66a1ac
We might add new values to dynamic_partition_info.txt, so some kind of
generic fallback mechanism is needed. If keys are different, we need to
decided on a case-by-case basis which side takes precedence. For
example, Virtual AB requires vendor support, so vendor side takes
precedence. VABC on T+ devices are implemented entirely in system, so
system/framework side takes precedence.
Bug: 230876542
Test: th
Change-Id: I67747368547d3ef3e29ad64f8f818ef4c5896246
add_img_to_target_files.py will re-make image and build a new
signed vbmeta.img with new images in target-file.
We need to add vendor_kernel_boot into AVB signing list and make it
support that image.
Test: make dist -j110 and check vbmeta.img
Bug: 214409109
Signed-off-by: Lucas Wei <lucaswei@google.com>
Change-Id: Id07433f3dc33f95e2edd49de890f1e098cb9ef31
GKI targets, e.g., `gki_arm64` or `gki_x86_64` have only
boot-*.img files without a ramdisk. It has no other images,
e.g., init_boot.img, system.img, etc.
The current build system assumes that the ramdisk is either
in a boot.img or in a init_boot.img, which is not true for
those GKI targets. Adding a new flag to support building
boot-*.img without a ramdisk while not building an init_boot.img.
Bug: 220834917
Test: build and `unpack_bootimg`
Change-Id: I789343c3e3d9ff0c36c0e19680a9792bd31a1c9f
Right now they're named ro.bootimage.* which is really confusing.
Change it so that if generic ramdisk is in boot.img, then name the
properties as ro.bootimage.*, otherwise if generic ramdisk is in
init_boot.img, then name the prop as ro.init_boot.*.
Bug: 211823573
Test: m dist
Change-Id: Icff3c7be04b9a3c6490fbf49ba4f05565b029890
gki_arm64 is a special target to build multiple GKI boot-*.img
files. It has neither system.img nor vbmeta.img. Skips checking
system.img and vbmeta.img in this case.
Bug: 223078405
Test: validate_target_files signed-gki_arm64-target_files-8277156.zip
Change-Id: I9d3c102c11787cf70c1e1fc8e6a552f8fd19ff72
Companion change of I143680b1cab50a6915df56c8273f8741beaf1180.
Basically does the same thing as the other change.
Bug: 211741246
Test: m dist
Test: ./boot_signature_info.sh boot-5.10.img
Change-Id: I40c4d5866c74a9a2d525f9455969b8a71f22bdbb
Some mountpoints are only available at device runtime, so ignore import
failures in release tools.
Bug: 214322896
Test: th
Change-Id: I1e8c4fb08c6b825fa19c0f0e019dfaf2b03f2927
boot.img and init_boot.img get their build properties from the platform
ramdisk properties.
Test: "m dist" then drop the OTA zip file into
android.github.io/analyseOTA to verify init_boot is present.
Bug: 211823573
Change-Id: I411e62bf0fc3cb49d4b12dadb52d9820fca041fa
zip -d doesn't work when no entry names specified. So if the list of
entry is empty, just skip calling zip -d.
Bug: 218438888
Test: th
Change-Id: Ie4419b9d6fdc2780255c6f12f9f4d35e5c0a7b26
