The mq-deadline I/O scheduler is more stable than BFQ. Additionally, the
boot time is shorter with the mq-deadline I/O scheduler. It only changes the scheduler on new kernel.
Bug: 186902601
Test: forrest build
Change-Id: I5d336b3f146ca673c83c03304773d0ccf4c01e77
Signed-off-by: Ocean Chen <oceanchen@google.com>
Set the variable is true when USE_LASSEN_OEMHOOK is true and
BOARD_WITHOUT_RADIO is not true.
Flag: EXEMPT bugfix
Bug: 325857852
Test: m ConnectivityMonitor
Change-Id: I10d1b9b198db960d1d6af0a69044907c8043c4e6
Pixel has removed unnecessary allocation padding, and we now pass
dEQP with a maximum physical core index setting of 4, which allows
the driver to utilize 4 GPU cores (1111b) in the extreme tests.
Bug: 221255664
Test: L10 atest CtsDeqpTestCases -- --module-arg 'CtsDeqpTestCases:include-filter:dEQP-VK.protected_memory.*'
Flag: EXEMPT bugfix
Change-Id: Iee15ed0d929f648ad45f7d5fe708e8cb14296e6f
(This reverts commit 9922a0fc12, ag/26704327.)
This requires that prebuilts on `main` have been updated to include the
April updates to the Secretkeeper TA (which includes fixes needed to
cope with larger and more realistic DICE chains). Relevant devices:
- pantah
- felix
- lynx
- tangorpro
Bug: 291228655
Bug: 306364873
Bug: 341708664
Test: VtsSecretkeeperTargetTest on cheetah
Flag: build.RELEASE_AVF_ENABLE_LLPVM_CHANGES
Change-Id: I251a14221915fa8b2aaf3f1611c8c9f698596552
This commit establishes a soong config variable `sim_count` for
easier converting Android.mk to Android.bp when dealing with multiple
SIMs.
Bug: 343823190
Test: presubmit
Change-Id: Idbd52236b95e8d69516226a71a848d591f181d94
The change disables the dumpstate strict-run flag in gs201 devices, which is enabled by default. Disabling this flag increases the timeout for capturing incident report to 120 seconds, from the default 20 seconds. This is done because 20s isn't enough to capture incident report proto in some cases, especially when device is throttling.
Bug: 330400782
Test: Booted a gs201 device with the change. Then manually thermal throttled the device and captured bugreports. In some tests Incident report took a lot more more than 20s, and in all these cases it was succesfully captured.
Change-Id: I177bd4aebc550ade855a8abf46ce261eea94d352
Add the flag "use_lassen_modem" for bp to do the same checking like
"USES_LASSEN_MODEM" in mk.
Bug: 330654744
Test: 1. lunch aosp_cheetah-trunk_staging-userdebug
2. m RilConfigService
Change-Id: Icb99b639837237b2afeea858de14669d583bf385
Revert submission 26384060-c2aidl-pixel-enable
Reason for revert: Culprit for b/335892284. Will be verifying through ABTD for confirmation and before submitting the revert.
Reverted changes: /q/submissionid:26384060-c2aidl-pixel-enable
Change-Id: I4acee85037309e699e82c8f378b72381be7de316
Revert submission 26467635-feature_cs
Reason for revert: b/329350496 - Introducing new config with `android.*` namespace is not allowed in 24Q2.
Reverted changes: /q/submissionid:26467635-feature_cs
Change-Id: I36c1408c09a3fa73e2958216e996a8d1e0d94487
Revert submission 26348985-bb-sepolicy-poc
Reason for revert: Droidfood blocking bug: b/327991669
Bug: 327991669
Reverted changes: /q/submissionid:26348985-bb-sepolicy-poc
(cherry picked from https://googleplex-android-review.googlesource.com/q/commit:3f5c12abada31befa0d9f869bf6628d06ee05aa4)
Merged-In: Id77188dd9f6f5acaed1506a9489f5d6f9874dde9
Change-Id: Id77188dd9f6f5acaed1506a9489f5d6f9874dde9
24D1-dev is based on 24Q2-release. Therefore, we merged this CL to 24D1-dev.
This change includes tachyon targets for edgetpu in the final image
TESTED:
Create a private build with this change and verified that the tachyon targets were present after flashing on a local device.
Bug: 307591410
Change-Id: Id584ee4ad626e28b7093e6df9a14c3e19dbe51ea
This reverts commit a929e8bc02.
Reason for revert: disable until gs201/main prebuilts get updated
The gs201/main prebuilts are a couple of weeks old, which means that some heap size fixes for the Secretkeeper TA are not present for these devices.
This in turn blocks:
- aosp/2997056: re-enable SK in pVMs
- aosp/3000947: add large DICE chain VTS tests
So revert for now, so we can submit those two CLs and have Secretkeeper test coverage happening on shusky.
Then when gs201/main prebuilts are updated to something more recent (i.e. that includes aosp/3005090), we can re-enable the HAL service.
Bug: 291228655
Bug: 306364873
Change-Id: I543310092ad6430a2dee30ae0b950fadbe869e15
The prebuilts on `main` have been updated to include:
- the Secretkeeper TA in Trusty (in ab/11346410)
- the ABL changes to retrieve the Secretkeeper identity pub key.
for the relevant devices:
- pantah (http://ag/26396674 to Trusty ab/11498491)
- felix (http://ag/26385511 to Trusty ab/11498491)
- lynx (http://ag/26395689 to Trusty ab/11498491)
- tangorpro (http://ag/26396492 to Trusty ab/11498491)
So enable the corresponding HAL service for Secretkeeper when the
relevant build flag is enabled.
Bug: 291228655
Bug: 306364873
Test: VtsSecretkeeperTargetTest on cheetah
Change-Id: I42804579ab4ff5a8ac0edd48fead8631f580dc37
Test: 1) complile and flash a gs201 device, passed PTS and CHQTS;
2) checked logs to make sure no sepolicy violations
Bug: 247124878
Change-Id: I0b0af38e89c13d645406bb1d30786bb25a74d963
Make the EdgeTPU targets defined in <soc>/proprietary visible by product packages included in edgetpu.mk.
Bug: 323601370
Tested:
No functionality change. Edgetpu services still work.
Change-Id: Ib3835ee3976dcfa86df999d0a32398bf8ec9a2cb
This reverts commit 2371ea7920.
Use the same i2c device paths across all kernel versions.
Bug: 323447554
Change-Id: I0f970347a09e01f272a54fc7c940ea576a9e510b
Signed-off-by: Jacky Liu <qsliu@google.com>
Enable dynamic color mode to reduce the power overhead of color space
conversion when GPU composition exists.
Bug: 315787778
Test: check "dumpsys SurfaceFlinger" and state of display color service
when playing SDR/HDR video
Change-Id: Ie9e87091385ab5247141bf956fbaba3275cc6811
