| Age | Commit message (Collapse) | Author |
|---|
|
Bug: 65455724
Test: make -j30
Change-Id: I03a22e10d92ea53bb2e1631e4203a186e42fa265
|
|
This reverts commit 05cd77c494aa7042fca7978e1b934855598b765e.
Bug: 65455724
Test: make -j30
Change-Id: Id531101cc35b661f02291eed0c044e96f5a91a12
|
|
This is the proper fix for the revert in ag/3048738.
Bug: 65455724
Test: lunch full-eng && make -j30
Change-Id: I88d37fca2d1f71a46595dd74b5fd6749fa4b477d
|
|
This broke full-eng / docs / m_e_arm-eng builds.
Test: lunch full-eng && m-j8
Change-Id: I79b0176738e4de35cca25e9618e64aa2bd4c4316
|
|
This CLs reduces BufferHub CPU consumption by adding asynchronous
state transition so that out-of-process VR composition can run on 2016
pixel devices smoothly. In addition, this CL addresses a couple corner
cases in the existing bufferhub logic, which fixes various blackscreen
issues.
1/ Tracks buffer transition states (gained, posted, acquired, released)
from the client side via atomic shared memory and adds
PostAsync/AcquireAsync/ReleaseAsync/GainAsync with metadata and
fence support.
2/ Adds dequeue order guarantee for buffers enqueued with
dvrWriteBufferQueuePostBuffer.
3/ Synchronous BuffeHub operations are still supported.
4/ Bump up the bufferhubd's soft limit of open file descriptor.
5/ Handle orphaned consumer in acquired state. This is a corner case
that consumer process goes aways (most likely due to a crash) leaving
buffer stuck in acquired state with inconsistent buffer state.
6/ Fixes a race condition for released buffer to be Gain'ed and
Acquire'd when a new consumer is created in released state.
7/ Improve silent consumer queue efficiency: Silent queues no longer
import buffers or receive signals about new buffers and they are
limited to only spawning other consumers and notifications about
producers hanging up.
8/ Modify PDX/UDS channel event signaling to work around epoll
behavior. PDX UDS uses a combination of an eventfd and an epoll set
to simulate the original PDX transport channel events. An odd
behavior discovered in the kernel implementation of epoll was found
that causes the epoll fd to "unsignal" itself whenever epoll_wait()
is called on it, regardless of whether it should still be
pending. This breaks the edge triggerd behavior in nested epoll sets
that channel events depend on. Since this is unlikely to ever be
fixed in the kernel we work around the behavior by using the epoll
set only as a logical OR of two eventfds and never calling
epoll_wait() on it. When polling is required we use regluar poll()
with the eventfds and data fd to avoid the bad behavior in
epoll_wait().
9/ Keep reading data after PDX hangup signal. UDS will signal hangup
when the other end of the socket closes. However, data could still be
in the kerenl buffer and should be consumed. Fix an issue where the
service misses an impulse sent right before the socket is closed.
Bug: 65455724
Bug: 65458354
Bug: 65458312
Bug: 64027135
Bug: 67424527
Test: libpdx_uds_tests
bufferhub_tests
buffer_hub_queue-test
buffer_hub_queue_producer-test
dvr_api-test
Change-Id: Id07db1f206ccf4e06f7ee3c671193334408971ca
|
|
The callback functions are set nullptr in ComposerClient::~ComposerClient()
./composer/2.1/default/ComposerClient.cpp:160: mHal.enableCallback(false)
However, the callbacks are still used in hwc2on1adaptor, it will cause segment fault.
Bug: 64650599
Test: adb shell stop
Change-Id: I92e6a6b42418e987a7ead041f2d95c194b6a16e8
Signed-off-by: Randy Xu <randy.xu@intel.com>
Signed-off-by: Qiming Shi <qiming.shi@intel.com>
Signed-off-by: Yong Yao <yong.yao@intel.com>
Signed-off-by: Gaurav Sarode <sgaurav@google.com>
|
|
|
|
|
|
into oc-mr1-dev
|
|
If we were unlucky with the timing when switching to vr flinger we would
sometimes deadlock. The vr dispatch thread would request the display
from surface flinger, locking mStateLock in the process. mStateLock was
being held by the surface flinger main thread, which was processing a
previous request to switch to vr flinger. The main thread was trying to
connect to the vr hardware composer service, which sent a request to the
vr dispatch thread as part of its initialization, leading to the
deadlock.
The deadlock is easily fixed by posting a message to request the vr
flinger switch to the surface flinger main thread, instead of doing it
on the vr dispatch thread.
Bug: 66916578
Test: Confirmed I can still get into/out of vr flinger on
Marlin. There's no longer code to acquire mStateLock in the vr dispatch
thread, so the deadlock is no longer possible.
Change-Id: I3de5476f698ed798f6b9afe927cc733f0f38c56e
|
|
HWC supports caching buffers for layers using "slot" assignments.
Use this in VrFlinger to avoid importing a buffer handle every
frame. The avoids periodic stalls we observe in the ion driver
when mapping a buffer into the HWC address space.
Bug: 66459419
Test: Observe systraces no longer have MapBuffer in HWC in steady
state; system does not drop frames.
Change-Id: Iba4161b33561322bfbccbfafe600b432a6fa7c44
|
|
Discovered that FP16 and 1010102 may not be supported
for Swapchain. Since we don't have a query mechanism
defined, limit P3 to 8888 which must be supported.
Bug: 66706404
Test: adb -d shell am start -n com.drawelements.deqp/android.app.NativeActivity \
-e cmdLine '"deqp --deqp-case=dEQP-VK.wsi.android.colorspace \
--deqp-log-filename=/sdcard/dEQP-Log.qpa"'
Change-Id: If537c631cd40610f6a62c437f8a7aff70f742fff
|
|
Can verify using this command:
adb shell dumpsys package | grep picture_in_picture
Bug: 65602088
Test: manual
Change-Id: I0e91cb63ce9d4b3ae059afc91d76a36ffc6a0a53
|
|
unchanged" into oc-mr1-dev
|
|
|
|
The function setVisibleRegions is always called as part of composition,
shouldn't mark geometry changed if there are no changes in the visible
regions.
Bug: 65019874
Test: run cts -m CtsViewTestCases -t android.view.cts.DisplayRefreshRateTest#testRefreshRate
Change-Id: I3f93431002fc4b691c799f4ae91aa1488849e4a6
|
|
oc-mr1-dev
|
|
Without this, PIP will break a bunch of apps since its frame fills the
whole screen, even though the application is restricted to its
stackbounds (which are the size of the touch region, not the frame
region).
Bug: 65606046
Test: None
Change-Id: Id7035dace30db37c5a58c66339d0749550a362e5
|
|
add getInstallerForPackage() and getVersionCodeForPackage()
to support native media.metrics needs.
Bug: 65027506
Test: watch logs for various package combinations
Change-Id: I6f3c551e8f547ab4151ef2400db4e100a6f53abf
|
|
We see two sources of scheduler jank when waiting for VSYNC:
- A kernfs issue that wakes up threads using a normal priority
work queue that may be delayed or have other work on it.
- The VSYNC callback from HWC is handled by a normal priority
HwBinder thread that may be delayed by other work.
Change the VrFlinger frame post thread to use an absolute timer-
based dead reckoning loop. VSYNC timestamps from the display
driver are reliable, even if the delivery of the value takes time.
Predict the VSYNC time into the future based on the last known
VSYNC time. If we see that VSYNC has not been signaled by the time
we need to post a frame to HWC we assume that the driver and/or
HWC was delayed so much that the previous frame is still pending
and skip the upcoming frame to avoid double stuffing the driver.
Bug: 65064949
Test: Extensive system tests and systraces. See bug for details.
Change-Id: Iae6c4173b8eac1d179adc3fc8004d3d475b3f156
|
|
|
|
This reverts commit 864c07077b0283ecf9bd36a0381951a19ec8c3df.
No special behavior for tests.
Reason for revert: b/65603915
Bug: 64776708
Change-Id: Ib64d4f6ef5ea59afa1d3c408c792685a3d025c0a
|
|
|
|
oc-mr1-dev
|
|
|
|
We sent up to 3 separate send requests over a socket. Now combine
then into one sendmgs() to improve performance and thread scheduling.
This improves PDX/UDS performance by up to 2x in some cases.
Bug: 65379030
Test: Sailfish works normally. Was able to run VR apps.
Change-Id: I00734d2c3f06d7d42b089569a8cc11c2273a1b7f
|
|
Do not validate the package path when reconciling secondary dex files. If
the file does not exist we cannot resolve the system sym links (e.g.
/data/user/0 -> /data/data) and the validation will fail leaving oat files
behind.
Bug: 64460009
Test: adb shell cmd package reconcile-secondary-dex-files
com.google.android.googlequicksearchbox (after removing some files)
adb shell cmd package compile -m speed --secondary-dex
com.google.android.googlequicksearchbox
adb shell /data/nativetest64/installd_utils_test/installd_utils_test
Change-Id: I9734ad18a579d44088180326661d8cf8288e90be
|
|
android.hardware.test.baz@1.0.so is installed to /data/nativetest(64),
and no longer to /system/lib(64) or /system/lib(64)/vndk, because the
interface library is only for testing.
Set runpath so that lshal_test can find the interface library in the
non-standard directory.
Bug: 65603915
Test: m -j lshal_test
adb push out/target/product/<name>/data/nativetest* /data
adb shell /data/nativetest/lshal_test/lshal_test
Change-Id: I854b08acb6442990ed369bfbc517ad1a93c94584
|
|
|
|
|
|
Adding new APIs:
dvrWriteBufferQueueGainBuffer
dvrWriteBufferQueuePostBuffer
dvrReadBufferQueueAcquireBuffer
dvrReadBufferQueueReleaseBuffer
Those new APIs streamlines DVR buffer operations:
1/ It allows static Dvr{Read,Write}Buffer objects to be statically
allocated with the queue.
2/ Symentric Gain/Post on WriteBufferQueue and Acquire/Release on
ReadBufferQueue.
3/ Use fixed buffer metadata.
4/ This is also a prerequisite to using shared memory based metadata and
async buffer IPC.
More details and discussion at: go/gvr-async-bufferhub
Special note regarding DVR API order in this CL: The new buffer queue
API entries are inserted before new PoseClient API as the PoseClient is
still blocked from being released and the new BufferQueue APIs are
expected to be released sooner than that.
Bug: 65455724
Bug: 65468551
Bug: 65458332
Bug: 65165821
Test: dvr_api-test
Change-Id: Iae8e7787d696d72ebf7457df9f98f4467cd20932
|
|
Avoid unnecessary tear-down and bring up of HWC client in standalone
devices. This saves time during screen-on because standalone devices
only use VrFlinger mode.
Remove dead code that used to check the panel driver for missed frames.
No drivers expose the sysfs node that provided this function anymore.
Test: Manual testing. Observe stable DON/DOFF behavior.
Bug: 65248224
Change-Id: Id9ebb76982621848d97792496a09b6da8c4e5928
|
|
|
|
oc-mr1-dev
|
|
Unlike other functions, executeCommands both sends and receives
handles. It can fail for reasons such as out-of-fd. We want to
protect SurfaceFlinger against out-of-fd.
Bug: 65449888
Test: manual
Change-Id: Id9420dbd6d0a943737c90bd3a1ae0df3805a4cb0
|
|
Setting up the VR hardware composer was causing a call to
setLayerCompositionType with an invalid composition type. This removes the
call, since setting the layer composition type is already handled in
Layer::Prepare().
Bug: 63633934
Test: Start VR mode, check that the logcat errors are no longer present.
Change-Id: I94f9a253bce9c5c25be7927d0003adbb9409ab36
|
|
|
|
Bug: 63683612
Test: Manual. Ran O2 on Marlin, tested image streaming manually.
Since DvrPoseRawDataType enum will be part of the public DVR API,
explicitly define the size of the enum.
Change-Id: I3900470ccb3ca7a2c4e7ed8e7f5586c60486d9e6
|
|
|
|
am: 047fcbf671
Change-Id: I17bac0d425207b1edc0ffbe997953502aae324e0
|
|
am: 6afce442ea
Change-Id: I7933009ee4c64025245373ee6b5ad1b0d77291e9
|
|
|
|
In a few cases we were accessing mHwc off the main thread without
acquiring mStateLock, resulting in crashes and other incorrect
behavior. This CL adds the missing locks. Since the locking semantics
are somewhat hard to understand, also add a clarifying comment to the
mHwc member declaration in SurfaceFlinger.h.
Bug: 64586546
Test: I manually tested normal surface flinger operation and vr behavior
on different devices, and confirmed everything looks fine. The crashes
we saw that were caused by these mHwc race conditions are hard to
reproduce, so I couldn't empirically verify this fixes the crash. I'm
relying on manual code inspection to confirm the issue is in fact fixed.
Regarding performance, I added profiling code (not part of this CL) to
check for lock contention and hold times with the newly added locks. I
confirmed that contention is low, so these calls shouldn't be
significantly slower as a result of adding the locks. The time spent
holding these new locks is also low, except for getDisplayColorModes(),
which makes a call to hardware composer service and can in some cases
take over a millisecond. That function is called so rarely though, only
once at boot, or twice at boot after a fresh flash, that it's not worth
optimizing.
Change-Id: I3854779c12a61983aaaecddb9f6316f218e519e3
|
|
|
|
READ of size 4 at 0x0079b1ac3a20 thread T121 (InputReader)
Viewport [1] to add: virtual:android:277f1a09-b88d-4d1e-8716-796f114d080b
#0 0x79b8820d53 in android::InputReaderConfiguration::getDisplayViewport(android::ViewportType, android::String8 const*, android::DisplayViewport*) const frameworks/native/services/inputflinger/InputReader.cpp:246:39
#1 0x79b8843fab in android::TouchInputMapper::configureSurface(long, bool*) frameworks/native/services/inputflinger/InputReader.cpp:3472:22
#2 0x79b88411b3 in android::TouchInputMapper::configure(long, android::InputReaderConfiguration const*, unsigned int) frameworks/native/services/inputflinger/InputReader.cpp:3248:9
#3 0x79b8825ff3 in android::InputDevice::configure(long, android::InputReaderConfiguration const*, unsigned int) frameworks/native/services/inputflinger/InputReader.cpp:1126:21
#4 0x79b8822653 in android::InputReader::refreshConfigurationLocked(unsigned int) frameworks/native/services/inputflinger/InputReader.cpp:612:25
#5 0x79b88239fb in android::InputReader::loopOnce() frameworks/native/services/inputflinger/InputReader.cpp:343:13
#6 0x79b882d507 in android::InputReaderThread::threadLoop() frameworks/native/services/inputflinger/InputReader.cpp:1004:14
#7 0x79d79e6113 in android::Thread::_threadLoop(void*) system/core/libutils/Threads.cpp:747:28
#8 0x79da914bbb in android::AndroidRuntime::javaThreadShell(void*) frameworks/base/core/jni/AndroidRuntime.cpp:1199:14
#9 0x79da20f31b in __pthread_start(void*) bionic/libc/bionic/pthread_create.cpp:214:18
#10 0x79da1c6dff in __start_thread bionic/libc/bionic/clone.cpp:47:16
Bug: 65378805
Test: asan build
Change-Id: I5a65cb007fe4879ce9635572014a568583a2ef4a
|
|
|
|
oc-mr1-dev
|
|
|
|
Bug: 63683612
Test: Manually tested through VrCore cl/164799996.
To retrieve raw data from Tango, added below api's to dvr_api:
- dvrPoseClientGetDataReader
- dvrPoseClientDataCapture
- dvrPoseClientDataReaderDestroy
Note: These changes have already been reviewed in branch
oc-dr1-daydream-dev branch. See change 2729572.
Changes tested on Pixel XL with and without 02 rendering path. Daydream
Home rendering in VR Mode correctly. Pixel functioning as normal.
Change-Id: Ia934d6a1a0b89edfd4408dde19d38e757d44f785
|
|
oc-mr1-dev
|
