* when duration = 3000, it may be too short for a app to complete the launch process
* follows pixel tablet launch boost value(5000) and extend to 6000 as sdm660 can't follow up gs201 speed
* this value is only working on the "first cold start" for an app, so make it larger is not affecting the whole battery life but just speed up the launch process
Signed-off-by: pix106 <sbordenave@gmail.com>
* schedtune is conflict with uclamp
* CONFIG_CGROUP_FREEZER is enabled
* CONFIG_UCLAMP_TASK is backported and enabled
Signed-off-by: pix106 <sbordenave@gmail.com>
Change-Id: I42170d51a517170b58d532addd9c38496e43457c
jasmine_sprout: Remove QTI perfd
Bug: 69270928
Test: Build and boot device and test camera
Change-Id: I87b283206f462fb5c8ec9cdd303ae2934fe9bfc4
jasmine_sprout: Declare BOARD_VENDOR
* Needed for xiaomi-perfmgr power AIDL guards
Change-Id: Idbb48e0f9b01090a456ad5d3f2d51f4dfab55672
jasmine_sprout: Create dummy libqti-perfd-client
* proprietary perfd blobs can finally be nuked without breaking goodix
* we could even map the functions to use libperfmgr powerhints in the future
[SebaUbuntu]: Cleanup Android.bp and add copyright header
Change-Id: I124652f3041761966a3e3bd97c757fecc39cc5fb
jasmine_sprout: libqti-perfd-client: Remove namespace declaration
It's pointless when using extern "C".
Change-Id: Ibdf9f06a70aa3a75687b33781c78cf2172bb334d
jasmine_sprout: libqti-perfd-client: return a dummy value
Return a positive integer for perf lock acquire and release so
that Goodix/FPC fingerprint sensor blobs will not complain.
Goodix:
E [goodixHAL][gf_hal_milan_f_series]: goodix_perf_lock_acquire: Failed to acquire perf lock, err: 0
E [goodixHAL][gf_hal_milan_f_series]: goodix_perf_lock_release: Perf lock release error 0
FPC:
E fpc_tac : fpc_perf_lock_acquire: Incorrect params, Failed to acquire perf lock, err
E fpc_tac : fpc_perf_lock_release: Perf lock release error 0
Signed-off-by: Chenyang Zhong <zhongcy95@gmail.com>
Change-Id: I861672e9a738c2204755d802670f4b28b662f286
jasmine_sprout: libqti-perfd-client: Move to C
* Why bothering with C++ mangling when we can just build it as a standard C library?
Change-Id: I45ea977edf7ea7fab6fece76f3049654a8d24c5d
jasmine_sprout: Add powerhint.json jasmine
Change-Id: If270fc906fd833bb6b0d3b00621c0b079346e47d
jasmine_sprout: rootdir: Initial libperfmgr init config
* From coral, heavily stripped down
Change-Id: Ib6846e4f4f23baf56910d2330e26846b99830218
jasmine_sprout: sepolicy: Initial libperfmgr sepolicy
Change-Id: I3d0a259bc89b4ecd0bf1632199172710d98230e0
jasmine_sprout: Don't explicitly build older minor version HIDL libs
Change-Id: Ib0caa49054a79b1f01ffc751826286a29c642e06
jasmine_sprout: Remove QTI perf reminants
jasmine_sprout: Add back ro.vendor.extension_library definition
* We aren't using QCOM's perfd but our blobs still check for this.
* To avoid unnecessary errors, let's just add it back.
Change-Id: Icc44b330f21a658cfa8ded691a6628d8f62c3649
Signed-off-by: pix106 <sbordenave@gmail.com>
* from android-13.0.0_r3
* Remove audio hints while at it
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
The ndk_platform backend will soon be deprecated because the ndk backend
can serve the same purpose. This is to eliminate the confusion about
having two variants (ndk and ndk_platform) for the same ndk backend.
Bug: 161456198
Test: m
Merged-In: I14a1c57bd06f1f2aa52491f779c7030d4de03547
Change-Id: I14a1c57bd06f1f2aa52491f779c7030d4de03547
Signed-off-by: pix106 <sbordenave@gmail.com>
* Let interaction runs on the most efficient freq for both big and little cluster
Signed-off-by: Clarence K <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
This reverts commit 1229fc5a8473c7ea99e90c4ff0a6f6d3a05a8b14.
* Nuked the schedtune.prefer_high_cap in kernel so it is not useful anymore
* Partially reverts only as some are still useful
Signed-off-by: pix106 <sbordenave@gmail.com>
This reverts commit 1229fc5a8473c7ea99e90c4ff0a6f6d3a05a8b14.
* Nuked the schedtune.prefer_high_cap in kernel so it is not useful anymore
* Partially reverts only as some are still useful
Signed-off-by: pix106 <sbordenave@gmail.com>
* when tested with [1], even after exiting the app, spm was
still active in powerhal thereby keeping max freqs limited
and blocking boosts
* enable/disable it properly and end all other hints before
enabling, to fix this issue
[1] https://android.googlesource.com/platform/cts/+/refs/heads/master/hostsidetests/sustainedperf
Change-Id: I672195fc6f7aba91928e2ab6147a6a7cac51f459
Signed-off-by: chiru2000 <chiranth@m.ms.evolution-x.org>
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
* end all running boosts when battery saver is turned on,
and block all boosts until its turned off
Change-Id: I888fe8f3d6779c22e26c24781c492d5576cb5610
Signed-off-by: chiru2000 <chiranth@m.ms.evolution-x.org>
[clarencelol]
* This one integrated into powerhal, much better
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
* Google removed stune support and misc in A12
* It keep spams failed to write error
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
[clarencelol: It improves scrolling janks in my testing]
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
Energy aware feature control is previously done through debugfs,
which will be deprecated, so move the control to sysctl.
Bug: 141333728
Test: function works as expected
Change-Id: I431b3113aa6bcde48b4b7cf9b2097666b948ad80
[clarencelol: These stuffs improve performance significantly]
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
With the kernel patch to separate placement hinting part of
boost signal on placement into a prefer_high_cap hint knob,
TA will lose the boost placement hint.
This CL adds placement hinting back to TA when INTERACTION
is on.
Bug: 144451857
Test: Boot & run jank test.
Change-Id: I3dcb39804d5bbaff344fbdaf64548335a1fdf1ea
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
set the boost value to 1 for CAMERA_STREAMING
Test: pixel test suite
Bug: 138323419
Change-Id: I56d1019d0cb0d206af6629566b09dd4743246289
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
Add a boosting control for the camera-daemon
stune cgroup.
Bug: 136608042
Test: Check reset of boost on boot
Change-Id: If0e42d16058989c385c9592a7080662566450b1e
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
* Remove unnecessary nodes
* Re-adjust scaling frequency
* Disable AUDIO Related hints
* Set both "Flipendo" freq to 1401600
Signed-off-by: pix106 <sbordenave@gmail.com>
* Set it according to kernel value shown here 3f4fbeaa44
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
Clover has a different dt2w node
Change-Id: Ic4f2a132b8a102eb01edab86469eccd7dfcd6c84
Co-authored-by: pix106 <sbordenave@gmail.com>
Signed-off-by: pix106 <sbordenave@gmail.com>
Because perfmgr is a vendor process, it cannot adjust system priority
directly.
Bug: 162791243
Test: build and using emul temp/running burn8 to verify it
Signed-off-by: pix106 <sbordenave@gmail.com>
* 5s is just too much long plus battery drainer
Signed-off-by: DennySPB <dennyspb@gmail.com>
Change-Id: I7e037c45cd2296a8bbd67d6bd63c212c0ef2a805
Signed-off-by: Henrique Pereira <hlcpereira@pixelexperience.org>
Signed-off-by: pix106 <sbordenave@gmail.com>
* Re-adjust scaling frequency
* Set both "Flipendo" freq to 1401600
Signed-off-by: OdSazib <odsazib@gmail.com>
Change-Id: I4e89982ca17fbceada830a55958d1db7806787f7
* Needs Kernel side support
* This activates kernelspace battery saver via powerhal whenever
battery saver is enabled in userspace, thereby lowering power
consumption at kernel-level by disabling boosts and such
Signed-off-by: clarencelol <clarencekuiek@icloud.com>
Change-Id: I2623503db27d7518de519bcaa3f5af6ab83879d0
