Age | Commit message (Collapse) | Author |
|
d7cdfd6b28
Original change: https://googleplex-android-review.googlesource.com/c/platform/hardware/google/graphics/common/+/14060222
Change-Id: I68f56771a7ec819ad58fde65d9b2a5d3e1b06b1e
|
|
Added a IR compensation(IRC) control into
the hbm mode property. "0" stands for hbm off,
"1" stands for hbm on and irc on, "2" stands for hbm on
and irc off.
Bug: 177144573
test: check IRC OFF
Change-Id: I28b6d472355007404e1c90b74c4b315186e42f8a
|
|
Original change: https://googleplex-android-review.googlesource.com/c/platform/hardware/google/graphics/common/+/15443818
Change-Id: I14e267fbf67c79d94a371a65a79f1395f207512f
|
|
Skip the HBM dimming when full-screen HDR switch.
Bug: 194655895
test: check dimming when full-screen HDR switch
Change-Id: I036f1764f5edcc009ea856510dfb579998c87abd
|
|
sc-dev am: ebcda1828a
Original change: https://googleplex-android-review.googlesource.com/c/platform/hardware/google/graphics/common/+/15395844
Change-Id: Id1571025e216b426d419a03874391afe6462772c
|
|
|
|
The hbm on by atomic commit is only for synchronization
with dim sdr layer transition. Otherwise, hbm on goes to
sysfs patch.
Bug: 194538187
test: lhbm on/off after hbm on/off
Change-Id: I56ee971de54ca521e755a27a0dcb97887935671b
|
|
fa798da1d0
Original change: https://googleplex-android-review.googlesource.com/c/platform/hardware/google/graphics/common/+/15391110
Change-Id: I271a98dbd9e953fa30e94fbfc75969afeb309a06
|
|
|
|
Original change: https://googleplex-android-review.googlesource.com/c/platform/hardware/google/graphics/common/+/15389252
Change-Id: I9b987d49691c41b172ebf86a0333e2f4066b73a4
|
|
|
|
This reverts commit 46ed0bcd8cdc7f31b1016ccc8f5861d0f660e07e.
Bug: 191426007
Reason for revert: regression issue LHBM not work
Change-Id: I5225b3db371fb551ad060b1a4dde30af429925bc
|
|
M2M MPP tends to reuse destination buffers when conditions allow. To
avoid the wrong configurations in FBID cache, FBManager checks the
color format for buffer cache.
Bug: 194369138
Test: G2D can work correctly when intermediate color format is changed.
Change-Id: Ib78e714265368aca2255ce211c11102b27a66685
|
|
Original change: https://googleplex-android-review.googlesource.com/c/platform/hardware/google/graphics/common/+/15093523
Change-Id: Ie0aad2ef525c8009c299392cd22dff6607156a79
|
|
|
|
Only allow to set LHBM on when state ON and
max refresh rate. And force to turn off when
Doze mode.
Bug: 191426007
Test: check lhbm/refresh rate
Change-Id: Ibf884294272abc1af947a322b9694061e06d9834
|
|
Original change: https://googleplex-android-review.googlesource.com/c/platform/hardware/google/graphics/common/+/15259413
Change-Id: Ia8afb9ac1534a59343d848da46d74a36da64910f
|
|
Use hw dimming for smooth brightness transtion.
nomral: enable dimming
hbm: enable dimming only for hbm transition
none: disable dimming
Bug: 190806064
Test: dimming on/off
Change-Id: I5d8c167c2e46c0ea740d104cfc3bcefe20717918
|
|
Original change: https://googleplex-android-review.googlesource.com/c/platform/hardware/google/graphics/common/+/15273538
Change-Id: I98626ecac8bf0d82ebf7201c565ebf8612ba8384
|
|
Some color mode related codes are added.
It handles color mode configuration which device is supported.
Bug: 192989878
Signed-off-by: YongWook Shin <yongwook.shin@samsung.com>
Change-Id: I697c5dcd329cbecf79640ea7762eee44f1cfd4bb
|
|
Dataspace check codes are added.
It checks dataspace configuration which device is supported.
Bug: 192989878
Signed-off-by: YongWook Shin <yongwook.shin@samsung.com>
Change-Id: I6cf1bdf9d0786457edb57ced7e5c33989ed2ae56
|
|
Bug: 188811178
Test: refresh rate switch
Change-Id: Ib8dc5efc25ebb0f29e3a7fcecb4fec7e27ff6a3c
|
|
* changes:
libhwc2.1: enable refresh rate mipi sync via drm prop
libhwc2.1: pass HDR full screen status to displaycolor
libhwc2.1: enable smooth display compensation
|
|
The secure buffers should be released since there is a size limitation
on secure buffer allocation. However, FBManager cached the secure
buffers from the client target longer since there was no signal to
destroy client buffers. The patch creates the new signal to destroy
FBIDs of secure frame buffers once there is no secure client target on
the screen.
Bug: 193041969
Test: atest GtsMediaTestCases
Test: secure buffers is freed once secure video playback is terminated
Change-Id: Ic28b7a540215cc8efffffa3bdf44bebcdb80e7ba
|
|
Enable refresh rate mipi sync via drm prop only when RR compensation
regamma exists.
Bug: 187763532
Change-Id: Ia368f656640bad32d33844a5ce3e32a8f70fe5e7
|
|
This patch refreshes frame or skipes frame update
when tui event is generated.
Bug: 157272869
Signed-off-by: HyunKyung Kim <hk310.kim@samsung.com>
Change-Id: I92affee60d02d8570378e2f2275ccd0a46b063ec
|
|
It sets full size of readback buffer as strides instead of
display resolution because buffer can have padding bytes.
Bug: 175422660
Signed-off-by: YongWook Shin <yongwook.shin@samsung.com>
Change-Id: Ie21810c5529a635f3e7c63f1b08d65f86c03e9e8
|
|
clearDisplay with needModeClear parameter clears ModeSet.
It forces to do Atomic Commit for ModeSet when next display is working.
Bug: 175422660
Signed-off-by: YongWook Shin <yongwook.shin@samsung.com>
Change-Id: I148b7e3cf1712b4a3b446d39d2da8049f09d30db
|
|
When readback buffer is required, readback properties should
be set when Atomic Commit for frame update is processing. When
this readback buffer is not required, readback properies also
should be reset.
Bug: 175422660
Signed-off-by: YongWook Shin <yongwook.shin@samsung.com>
Change-Id: I8bcef3d844ac09a05ebe6e9fec1dee9416fcd6a4
|
|
Error information of DrmModeAtomicReq could be overwritten
when drmModeAtomicCommit() returns error in the previous code.
This patch doesn't call destroyOldBlobs() in DrmModeAtomicReq::commit()
to fix this bug.
DrmModeAtomicReq destructor always calls destroyOldBlobs()
so we can remove it from DrmModeAtomicReq::commit().
Bug: 175422660
Signed-off-by: YongWook Shin <yongwook.shin@samsung.com>
Change-Id: I1a1ee3e33231c3ce2d9da8ea9803954fd4a2d78f
|
|
This patch disables DQE if readback is requested
except readback is requested by HWCService.
setDisplayColorSetting() call is skipped as well in
this case.
Bug: 175422660
Signed-off-by: HyunKyung Kim <hk310.kim@samsung.com>
Change-Id: I9332ae351834f9c7241163214cbd4f5d148827a3
|
|
into sc-dev
|
|
VsyncWorker may reset config state when a desird config is still
pending in main thread. So we should reset the config state only
when the state is REQUESTED.
Bug: 187151493
Test: swich refresh rate quickly, suspend/resume, AoD
Change-Id: I2c3186432e4aa93e59007de7700d360f5eecddd6
|
|
FBManager can cache too many fbIds with the exhaust of iommu space for
drmdecon. To avoid the system being unstable, FBManager will clean up
unused layers when the layer count exceeds the limitation.
Bug: 190541937
Test: Use test binary to simulate layer leakage
Change-Id: If3e99a4398cf64406f378cea35f266eadf20832b
|
|
|
|
Bug: 184768835
Bug: 185087460
test: test lhbm function call
Change-Id: I924d162713d95954855c706287f10f8cb696f763
|
|
SurfaceFlinger cache is allocating one new buffer everytime to cache
layers. Before the problem is fixed in the SurfaceFlinger side, the
patch cleans up the newly created fbIds when the cached size exceeds the
limitation.
Bug: 189203448
Test: Trigger SF cache in YouTube and Camera
Change-Id: If4f8c0920abeb3f5c6c8331ba8cb433ef75e0bf8
|
|
fbIds for solid color layers should be cached and freed properly to
avoid leaking. The patch supports fbId cache for solid color layer.
Bug: 189203448
Test: Stress test on YouTube
Change-Id: Ic910de7e5519ccb6c5cf4685f9d477e450303e22
|
|
The previous method to bind FBIds with handles was unstable because
the imported buffers were deleted without any awareness of underlying
HWC implementation. To make a robust mechanism, we bind FBIds with
unique buffer ids to avoid the collision in FBIds or handles.
Bug: 185958099
Test: Stress test on livewallpaper
Change-Id: I48c8f6cf044ef48fe2c0886ac9430e2313e4765f
|
|
* changes:
libhwc2.1: change the hbm mode and brightness update sequence
libhwc2.1: reset DimmingOn/HbmOn when power off
|
|
Since DisplayManager handles HBM for HDR, remove
the HDR case from HBM control in HWC.
Bug: 187673576
test: check HBM HDR use case
Change-Id: I6f346c5420e092513318f01244660e897f375635
|
|
Followed the hw panel sequence which updates the hbm mode
before brightness level.
Bug: 184120990
test: check the brightness change
Change-Id: I13724b805aa9a3d32e6e8b33be050067ddbc8129
|
|
The dimming/hbm of panel will be reset to off after power off.
It has to configure DimmingOn/HbmOn again when power on. Set
DimmingOn/HbmOn to false when power off.
Bug: 184120990
test: check the value of panel reg
Change-Id: Id8ce7a3aae8a91a6585eb36de22afefe9a17db69
|
|
Enable LHBM mipi/frame sync to compensate the luminance difference
Bug: 184096624
Change-Id: Ic84ca018933f042692e656a73632db56bc2968a6
|
|
LHBM has the frame sync synchronization requirement. Added the
drm property path for LHBM.
Bug: 185086850
test: check lhbm on/off
Change-Id: I9acda2e99ebf60a42bdc92da1b61047552bcd6d0
|
|
The way to trigger next frame (HWC2_CALLBACK_REFRESH) can not guarantee
the next frame arrvies timely. But the synchronization between mipi
command and frame content depends on each frame arrives on time.
HWC now handles the process in a synchronous way.
Bug: 184857029
Test: Unlock device with UDFPS global hbm solution.
Change-Id: Ie3c063fa81fba7cf88b2a5aae9ca6332772afb5c
|
|
Move color data update logic in one function.
Bug: 184857029
Test: Verify display color is correct in Natural mode.
Change-Id: I70ccd71cec562cab093e69375e72d2851684e81a
|
|
Driver is in fully ready status after this call.
Bug: 186446401
Test: Reboot and no failed reading panel serial in displaycolor log.
Test: Device could switch to 60/90/AOD mode.
Change-Id: I4edd4b263c0beacf33171773d90be71b30a16eee
|
|
FramebufferManager::cleanup sometimes took 100ms and more. Deferring the
action of drmModeRmFB from the critical session can improve the
performance in HWComposer.
Bug: 184845481
Test: stress test on FramebufferManager::cleanup
Test: Reboot test overnight
Test: Simulate the long latency in ~Framebuffer()
Change-Id: I506d77335768f0c2b97735440425cbe88789374b
|
|
A race condition is detected between getBuffer and cleanup where the
same element in mCachedBuffers will be deleted twice.
Bug: 170327170
Bug: 183912416
Test: stress test on FramebufferManager::cleanup
Change-Id: I4e1af64373982356ce4c4335be73dd06a4f88bae
|