Age | Commit message (Collapse) | Author |
|
* changes:
libcompositionengine: avoid vector<const T>
librenderengine: avoid vector<const T>
libdumputils: avoid set<const T>
|
|
|
|
|
|
These services were last present on the Pixel 3.
Bug: 234559097
Test: presubmit
Flag: EXEMPT dead code removal
Change-Id: I7bac2eeeb77bea8c048fcbe5f9ed872a039edc9c
|
|
Prevent TimeStats from getting stuck waiting on a stale time record
after multiple setBuffer/setPostTime occurred for the same layer within
the same commit.
For example:
V TimeStats: [109]-[1]-[SurfaceView[...](BLAST)#109]-PostTime[...]
V TimeStats: [109]-[2]-[SurfaceView[...](BLAST)#109]-PostTime[...]
... 62 frames later ...
V TimeStats: [109]-[64]-PresentFenceTime[...]
V TimeStats: [109]-flushAvailableRecordsToStatsLocked
V TimeStats: [109]-[1]-presentFence is still not received
Bug: 351133405
Test: run TvtsJankHostTestCases and compare the presentToPresent
histograms between the UI and video layers (missing lots of video
frames), or watch for
logcat -s TimeStats | grep "timeRecords is at its maximum size"
Change-Id: I3de44250d5eb919f58d60976d7c33bd8394c0f0c
|
|
A container of const T uses std::allocator<const T>, which was an
undocumented libc++ extension that has been removed.
See https://github.com/llvm/llvm-project/pull/96319.
Bug: 349681543
Test: m libcompositionengine
Change-Id: I359350a298fc9a59c0ca925a36f753ac3fb3b64e
|
|
Bug: 349735052
Add LOG_NDEBUG define in LayerLifecycleManager for debug.
Test: make & boot
Change-Id: I160fcaa6527aa120b2ba300d6a3d3eb6be649a8c
Signed-off-by: fengqian <fengqian@xiaomi.corp-partner.google.com>
|
|
Original change: https://android-review.googlesource.com/c/platform/frameworks/native/+/3073925
Change-Id: Id03328934cfc5424960812ba0c5f2d1d40b84216
Signed-off-by: Automerger Merge Worker <android-build-automerger-merge-worker@system.gserviceaccount.com>
|
|
|
|
aosp-main-future
|
|
Bug: 337098550
Merged-In: I8d975b81afd2cde1d543e8bc7e730a6e42393bec
Change-Id: I714982d2b0df34afc9f9729deb2636de5d0c7b64
|
|
data
Test: system boots with no display identification error logs
Bug: 220045335
Change-Id: I6efa1daf523b17ea56f4a4b9e842d663804e1c15
|
|
setPowerMode() is not called in HWC virtual display
so capabilities was not updated.
This change updates capabilities when HWC virtual display is created.
Bug: 329533475
Test: manual
Change-Id: Ib12a5e76cd9b1efb0dd14a6ad31983e41a0eeaba
Signed-off-by: HyunKyoung <hk310.kim@samsung.com>
|
|
Set the container field of aconfig flags
Test: printflags --format='{package}:{container}:' | grep "::"
Bug: 312769710
Flag: NONE
Change-Id: Ic5c3c7839cb4c34c8a97d841ccaeaeda0771cbcd
|
|
Bug: 328444881
Test: m
Change-Id: I13a8ff9a0c049a71ad8945d592fa53ad34e8139a
|
|
Restore some SF owners to certain test files to make reviewability
easier.
Bug: none
Test: presubmit
(cherry picked from https://googleplex-android-review.googlesource.com/q/commit:bb008aaccf30ef62773214cdfcfabb43aa33b8b2)
Merged-In: Ibe820409564b8bc330eedc4bf60ea17a84d012a4
Change-Id: Ibe820409564b8bc330eedc4bf60ea17a84d012a4
|
|
into main
|
|
This patch fixes the flaky notifyPowerBoostNotifiesTouchEvent test by
forcing all operations to happen in a deterministic order, avoiding
races.
Bug: 332875603
Test: atest libsurfaceflinger_unittest:DisplayTransactionTest
(cherry picked from https://googleplex-android-review.googlesource.com/q/commit:0fd347a02ea101c3557083ccb935342e42bc977c)
Merged-In: I882c8cce57f071ec0c4c4e59f1b00dcca090c650
Change-Id: I882c8cce57f071ec0c4c4e59f1b00dcca090c650
|
|
We may miss some state changes if a display state change comes between
processDisplayChangesLocked and commitTransactions. Fix this by grabbing
the state lock for the duration of display updates in commit.
Test: steps in bug
Bug: 330105711, 330103914, 328539539
Change-Id: I4798961551f78d75c45ead6dea5ebca895e5ef7d
|
|
Currently a lot of the SurfaceFlinger tests owned by WM automatically go to
the first person listed in the parent OWNERS file from CoGS. Extracting
tests with clear ownerships in this OWNERS file to reduce triage overhead.
Bug: 334170221
Test: N/A
Change-Id: I9728f07d4411bd9a55984992e0b520624c6be90c
|
|
* Remove chavi :'(
* Add sally
Change-Id: I12a329bd200391e25b0f0e7047b9c74a0cc4cf1b
|
|
|
|
Bug: http://b/289090109
Test: treehugger
Change-Id: I88711743d62687dfa480fa76fbb96763fdf03347
|
|
|
|
Protobuf 22.x no longer pulls in <set>, it needs to be
included explicitly.
Bug: 329747255
Test: presubmit
Change-Id: I8c0b219322b10a655d5687c506d1047ddb8aa99e
|
|
None of our libcs need this any more.
Change-Id: I83de62b66c11ed36db0faee746cc2a68708c6fc6
|
|
|
|
This reverts commit b43b58378679bf4ac8d0837047187d4509bc45d8.
Reason for revert: Appstartup regression found in b/327488235
Change-Id: I8d975b81afd2cde1d543e8bc7e730a6e42393bec
|
|
|
|
|
|
- Freeze the old flag file
- Create a new file, alphabetize and annotate
brackets to reduce merge conflicts
Test: presubmit
Change-Id: I6b6a1c81dea39f22eca5dc671e0cfedc7d139b25
|
|
Fix HighHint touch boost regarding frame rate override. The
touch boost logic in getFrameRateOverrides should mirror frame rate
selection in getRankedFrameRates. We decided earlier that HighHint is
only ignored when there are ExplicitDefault compatibility votes.
Bug: 326666127
Test: atest libsurfaceflinger_unittest
Test: jank test cases + manual
Change-Id: I54ac61f2a174b9a4dea7d67d8b196217b4f60921
|
|
Bug: 175635923
Test: m checkbuild
(cherry picked from https://googleplex-android-review.googlesource.com/q/commit:94a21e9ae2ea8162128fe392f743d523238b7f3e)
Merged-In: I9cef8d1b0306fab363e2fe4eda39782e21f24d8f
Change-Id: I9cef8d1b0306fab363e2fe4eda39782e21f24d8f
|
|
|
|
|
|
|
|
Add the concept of timeline freezing when switching render rate.
This allow us to change render rates in sync with the app and remain
jank free across render rate changes.
Bug: 326599221
Test: Run TouchLatency, change render rate and examine Perfetto trace
Change-Id: Ibc8026434c0c1d50138299da3cb110b317604e92
|
|
|
|
commit 4509966315ca9334d4637f8f364a34a12985ab13 accidently removed
the read only attribute from this flag.
Bug: 307674749
Bug: 273702768
Change-Id: I27ff11077226d5bc13731f4fd86dff0dc183ea47
|
|
To discern the source of the frame rate vote, this change
adds trace in LayerHistory when determining the layer vote.
Also, dumping information of game mode and game default frame rate
overrides in SurfaceFlinger dump for debuggin purposes.
Bug: 324125833
Test: atest LayerHistoryTest
Change-Id: Ice628b7f421f6402fbb1e33413966c503bf5eaec
|
|
protected buffer usage of output buffer of framebuffer is determined
by if any input layer contains protected buffer.
If all the protected layers are handled by HWC,then GPU will process
normal layers only. It means, GPU got normal buffer as input but need
output to protected buffer.It would be memory violation for ARM as it
required at least one normal buffer as input for such case
Cherry-picked from AOSP. Conflicts in the test file are uninteresting;
both branches simply added new tests.
- services/surfaceflinger/CompositionEngine/tests/OutputTest.cpp
Update OutputComposeSurfacesTest_HandlesProtectedContent to pass.
Bug: 307674749
Test: atest
libcompositionengine_test:OutputUpdateProtectedContentStateTest
Change-Id: I11d7c73c8eee4a46383516e8eb672827b26314e6
Merged-In: I11d7c73c8eee4a46383516e8eb672827b26314e6
|
|
|
|
Validate the GTE frame rate compatibility value, otherwise it does not
reach SF.
Bug: 327013543
Test: atest FrameRateUtilsTest
Test: atest CtsSurfaceControlTestsStaging
Change-Id: I3cb7fc7c9cbd32eb8db8e07718a89c3918b72a9f
|
|
|
|
Create helper functions to improve readability of what is scheduled on
the SurfaceFlinger main thread. This will allow for cleaner changes in
reducing the calls on the main thread for screenshots. Changes include
some renaming for better clarity.
Bug: b/294936197
Test: presubmit
Test: atest SurfaceFlinger_test
Change-Id: I729b62b3fc1919b5b048b7992b4aead336e87d7f
|
|
|
|
* changes:
[SF] Change VSyncDispatchTimerQueue::Schedule to return ScheduleResult
[SF] Enable vrr_timeout_hint_enabled
[SF] Send NotifyExpectedPresentHint at the transaction time
|
|
Hide the Render/LastComposite details in Scheduler.
Also, dispatch hotplug errors to both EventThreads for consistency.
Bug: 241285191
Bug: 241285945
Test: Hotplug still works.
Change-Id: I425d997274c0272a2bfa352c138750d65d008da5
|
|
|
|
This abstraction never made sense, and only incurs pointless locking and
hash map lookups. This is a step toward replacing the two EventThreads
with a single EventDispatcher.
Start the Scheduler's timers after initialization. This solves the same
bug as Ib2617b914145bc4180cc7ca27203c59dbd625c94, which avoided crashes
without fixing the root cause of the race.
In a few cases, initScheduler indirectly calls Scheduler functions that
attempt EventThread dispatch before EventThread creation, so skip those
EventThread accesses using hasEventThreads for now. Further refactoring
will untangle circular dependencies and remove hasEventThreads.
Bug: 241285191
Bug: 241285945
Test: presubmit
Change-Id: I89b93b47540ab24bb5675edc79370f38bc48114f
|