Retention is leading to increased rendering time, and has not been
measured to provide any power benefit. This Cl is to remove that.
Bug: 37287554
Test: muskie boot and check retention disabled after boot
Change-Id: Id6162070220b792958dce91758660c0d2aed0195
init.rc already set cpuset to all core during boot
Bug: 36576280
Test: marlin boot fast and checked cpuset during early boot
Change-Id: Ib91275361c8f8a1744f99b2da0b2f1769f31072c
EAS v1.2 uses different search ordering in the wakeup-path. Now cores 4
and 5 will be searched first for top-app, update the cpusets accordingly:
Littles Bigs
0 1 2 3 4 5 6 7
T T T T T T T T (top-app)
F F F F F F (foreground)
S S S S (system background)
B B (background)
Bug: 36367381
Change-Id: I7f70493ff45779d1c17f19465bf3eeeb8905e05b
Signed-off-by: Joel Fernandes <joelaf@google.com>
Recent CL make VR services hard-fail (crash) on failed attempt to
create PDX sockets. Wahoo-based devices don't have socket directories
under /dev/socket/pdx/... to create those sockets, so they now crash
at boot (which includes surfaceflinger). This prevents the device from
even booting. Add init script to create those socket directories.
Also, create necessary cpusets used by VR services, similar to what
was done for marlin/sailfish (see this commit in device/google/marlin:
1cc612c4ee646e8fc6ae69e8837b9b0d0b2119ce).
Bug: 36977281
Test: `lunch walleye-eng && m -j32 && flash` - build succeeds, device boots
Change-Id: Ic458ceabb16917bcf574f6d8e2de864d41a44d96
This adds the vibrator HAL that implements the new setAmplitude and
perform(Effect) API.
Test done: vibrator_hidl_hal_test
Bug: 36782452
Change-Id: If9988434277790becb469d4dd928e75f7e6af41a
Signed-off-by: David Lin <dtwlin@google.com>
1) separate fstab mount
2) launch bootanim early
3) boottime cpuset/io tune
4) parallel slow init operations: write sysfs and insmod
After CL:
Boot time saved 330ms, bootanim triggered early before data/ mounted
Bug: 36780513
Test: walleye boots
Change-Id: I02803179746710413f4b1e2372f3550cd95d1581
Test: flash build, confirm CHRE daemon starts, SLPI does not crash, no
SELinux denials, and context hub HAL connects to it and works
normally
Change-Id: I7af26548ed6e5b0544db621468a8cb8839f878ed
Test: NFC powers up without any denials
Bug: 36686703
Change-Id: Ib33a0042c5d03d2b9ee8a02dac143da9c8c216a9
Signed-off-by: Ruchi Kandoi<kandoiruchi@google.com>
1) device.mk: add charger_res_images to PRODUCT_PACKAGES
to use the default charger mode images
2) init.hardware.rc: configure charger service class
3) BoardConfig.mk: enable suspend in charger mode
Bug: 36046603
Change-Id: I3652a69784b8ecd4e670f26fdd66b7c68d578f9f
Signed-off-by: Yueyao Zhu <yueyao@google.com>
Cache is not used on the device using AB update. In that case, init just
waits
for timeout 5s.
[ 18.904216] init: Command 'mount_all fstab.${ro.hardware}' action=fs
(/init.taimen.rc:33) returned 1 took 767.808ms.
[ 23.911072] init: Command 'wait /dev/block/bootdevice/by-name/cache'
action=fs (/init.taimen.rc:40) returned -1 took 5006.36ms.
Test: Build and tested on a device
Change-Id: Iaee76961c98b7b7873e532a7003d560ebf6c4752
