The cancelling behavior in MTP has two behaviors.
1. cancel requests one by one
2. batch cancel all requests
This patch will set the property to true to enable
batch cancel all requests.
Bug: 181729410
Test: build pass
Change-Id: I8d92cc676ff7be8fd22a574b32653bafc028b801
SurfaceFlinger IPC improved with mid and big cores, relaxing cpuset will
help improve its efficiency. For render engine, we set uclamp.min to
skip scheduling on little cores.
FWIW, we let the RT scheduler figure out what is good for perf/power.
Bug: 170507315
Bug: 191054286
Test: Build and boot
Signed-off-by: Wei Wang <wvw@google.com>
Change-Id: I880cf549e33a3a2f629e8260366c4ce8c34c3ae6
Specify ownership to system and r/w permissions to user/group.
Allows dumpstate to access /dev/trusty-log0.
Bug: 188285071
Test: lshal debug android.hardware.dumpstate@1.1::IDumpstateDevice/default trusty
Change-Id: I935cb9dd11d1e0968ab0e2c9d7349de336502e7c
The cancelling behavior in MTP has two behaviors.
1. cancel requests one by one
2. batch cancel all requests
This patch will set the property to true to enable
batch cancel all requests.
Bug: 181729410
Test: build pass
Change-Id: Iad744e27e6f843ae22fc499efbb803be51e1b820
This reverts commit b6c4f766b748af81d502cf031c75b2c6c4c7cf09.
Reason for revert: switch to use ro.boot.hardware.sku for judgement
Bug: 188396079
Change-Id: I012b2d32d9b7245fd1d1be5a00c5e922a1f52d19
Give local_hbm_mode node sutiable permission to allow
hwcomposer to set local_hbm_mode to display driver.
Bug: 184768835
test: check file node permission
Change-Id: I2e5dcd87ba27a1ccf8de14a12472ba806833ed12
Bug: 172793258
Test: built and booted on oriole, tethering still works
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: Iff304938e57e28b3a62784fbd0b226a20e0f53cc
Bug: 188365095
Test: Local Test
/dev/thermal/tz-by-name/soc # ls -la
-rw-r--r-- 1 system system 4096 2021-05-26 17:25 mode
-rw-r--r-- 1 system system 4096 2021-05-26 17:25 trip_point_0_hyst
-rw-r--r-- 1 system system 4096 2021-05-26 17:25 trip_point_0_temp
/dev/thermal/tz-by-name/soc # cat policy
user_space
Signed-off-by: George Lee <geolee@google.com>
Change-Id: I9f9cffb29ad025872d77dc2e41d893a2fb2d375a
On userdebug & eng builds, change the owner/group of SJTAG sysfs files
that need to be written to 'system'. This became necessary after these
files were migrated out of DebugFS, as write access to the world is no
longer supported.
Bug: 188817349
Test: Ensure that the writable files in the following directories have
'system' as their owner & group:
/sys/devices/platform/sjtag_ap/interface/
/sys/devices/platform/sjtag_gsa/interface/
Signed-off-by: Peter Csaszar <pcsaszar@google.com>
Change-Id: I4420fccd92522a48f44b80bd0a19f4a20eb32442
Prefer high cap will choose the mid / high cores when possible with
consideration of energy efficiency.
Running task on high cores with low clock is more efficient than running
at little core with high clock.
This hint also addresses CTS issues with keeping latency_sensitive = 1.
(equivalent to prefer idle)
This hint adds:
1) Enable prefer_high_cap of camera vendor group
2) Set high_capacity_start_cpu to 4
4) Set MemFreq floor
5) Change camera-daemon/cpu.uclamp.latency_sensitive owner to root
Bug: 186564130
Test:
android.hardware.camera2.cts.RobustnessTest#testAeAndAfCausality (5/5)
android.hardware.cts.CameraTest#testPreviewCallback (5/5)
android.hardware.camera2.cts.CameraDeviceTest#testCameraDeviceRepeatingBurst (5/5)
android.hardware.cts.CameraTest#testFocusDistances (3/6). (3/6) without CL.
*****************************************************************
raven Camera CTS Test Results
*****************************************************************
Tests Passed: 217, Failed: 0, Not found: 0, Status unknown: 0
Change-Id: Ia2abb0eb9c57fa05c4db7bf1d02053e303301781
This reverts commit 3a0efffdbc722cb8a9684ce3c0080d2476c0abb8.
Reason for revert: This CL causes performance/frame drop issues which leads to more CTS flakiness.
Bug: 188462637
Bug: 185233206
Test: Camera CTS
Change-Id: I1796ba29da0e2c19874531bda8c21bc10e248055
Bug: 188110823
Test: Ensure that Dwell-Defend can trigger with 'setprop
vendor.battery.defender.threshold 61'
Change-Id: I4782e2c387b147fc48cf4134d27af8675964dc44
Bug: 188078957
Test: Verified using temporary log messages to confirm
file paths are formed correctly as otherwise to test queryMoistureDetectionStatus()
requires some mock circuits.
Signed-off-by: Manish Varma <varmam@google.com>
Change-Id: If5402d5999c77bf5a3431c70d800e50570b87c88
Since cfs_rq clamp feature uses tg's value, align it with the
vendor_group setting.
Bug: 170508405
Test: build pass
Signed-off-by: Wei Wang <wvw@google.com>
Change-Id: Ie0c896de2022f435be5ea53ae3cecf5d80d2c067
1) Enable prefer_high_cap of camera vendor group
2) Set high_capacity_start_cpu to 4
3) Set prefer_idle of camera vendor group to 0
4) Set MemFreq floor
5) Change camera-daemon/cpu.uclamp.latency_sensitive owner to root
Bug: 186564130
Test: Boot & check setting values
Change-Id: Ief0cae2aa9e4f84bed93f7de0bfb765e8a27009b
With vendor kernel support of getting rq cfs util in group base,
we could limit uclamp.max of background group to get more power benefit.
Bug: 170508405
Test: build pass
Change-Id: I997e653eb2790f5e3be4fbf3c8b1de356d787377
Right now we have many groups due to uclamp setup, and this would
potentially create interesting fairness between groups.
Ultimately we want only a system group and a background group for
cpu.shares, which would achieve after b/182509410 resolved.
Test: boot and check setting
Bug: 171740453
Change-Id: I8fa23268aa7ed02dcccc527abcec8556009aa9a7
This changes the owner of specific logbuffer device node
dev/logbuffer_btlpm and dev/logbuffer_tty16 to Bluetooth,
so Bluetooth HAL is able to access them in userdebug rom
Bug: 177794127
Test: Manually Test
Change-Id: I17db3e3b8df521946d4358d191695c09ac2c2c3a
This reverts commit 637fc9e8d727121276cf8dea2976ac52f8da8d15.
Reason for revert: Camera EBUF overflow observed after launching GCA.
Bug: 187760882
Change-Id: I2357d943f4b4d96ffa7a156671ac7c790640a5ed
uclamp value is calculated based on the full capacity 1024. Previously
camera-daemon uses 6 which would set the min to 61. This boosts the
camera thread util up and would have power penalty. Change the value
to 1 to have minimum boost.
Bug: 186564130
Test: cat /dev/cpuctl/camera-daemon/cpu.uclamp.min & check power
Change-Id: Ib8f5ae2091848da4ee98211aeae37612ae2b308a
With recently merged kernel change pa/1916082, down rate limit will be
adjusted dynamically. Without out this change, we would run into cases
that we too little down rate limit for little.
Bug: 186392111
Bug: 170511089
Test: boot and check rate limit
Signed-off-by: Wei Wang <wvw@google.com>
Change-Id: I78d9894301c9f4f888d33010c9fcca20d51d4b20
Bug: 187302501
Bug: 186636898
Test: boot and wifi function is ok
Test: p2p connection and throughput tests.
Change-Id: I599763e8c583f9566833183d4dbec13db1d040bc
This reverts commit 788012640b
Reason for revert: Caused throughput issue in p2p b/187302501
Bug: 187302501
Bug: 186636898
Test: boot and wifi function is ok
Test: p2p connection and throughput tests.
Change-Id: I70f4015040f10452acff697ee504b02cfb3c1e27
For dump these informations into bugreports and debugging
Bug: 186872139
Signed-off-by: Jenny Ho <hsiufangho@google.com>
Change-Id: I5e4efd83b1ae111a423cb13e1c35f31766f98252
The original edgetpu_service is now split into two, where the
original one stays under /system_ext/ and the new one,
edgetpu_vendor_service, now locates under /vendor/.
This change updates files including device.mk accordingly to
ensure these services are setup properly on P21.
Bug: 185432427
Test: Oriole + GCA
Change-Id: I5b0eff64f2f044a56bc14cf5e1d92ec3463560a9
With a scheduler change that refers to uclamp value of a rq, we need
to change uclamp min value for camera to get better performance.
Bug: 170648224
Test: check fps in video recording
Change-Id: I29e55706b49d5182935209d1d6f3552fb7f5d821
See http://shortn/_X7vYHND2xN for testing with various NN models
faster execution of NNAPI results in less execution time and
in turn, less total power consumed.
Bug: 183576817
Test: adb push'd rc file to a device
`cat /dev/cpuctl/nnapi-hal/cpu.uclamp.min` == 15.00
ran MLPerf, got better scores
Change-Id: I398c3741fc46958f95a9b608a27b52da47c69597
