Age | Commit message (Collapse) | Author |
|
Change-Id: Ib484346cff8ca194d43083c8a9c91cd03c57aa83
|
|
Cherry-pick from CL 3577634 on partner gerrit
When there's no buffer handle update for a given plan, current HWC implementation
will assume no content update and will not commit/flip the new content to display.
When the layer stack is changing (new layer added/removed...This's advertised by
SF through GEOMETRY_CHANGED flag), this assumption is no longer valid.
This patch will reset the current data buffer attached to the given plan when it's
being reclaimed. This should happen when rebuilding the new layer stack to be composed,
by the time we deinit the current layer list.
This way we can prevent black screen issue that may occur when a new activity's layer
keeps using the same buffer handle attached to the previous layer.
Change-Id: I4c24c963466db478633121dadcb6434208fc1b4a
Tracked-On: https://jira01.devtools.intel.com/browse/AW-7570
Signed-off-by: Zaghdoud Walidx <walidx.zaghdoud@intel.com>
|
|
208 snap-temp-L10800000088711883
Change-Id: I29c29bb66f4f5d1e100ffe1a7ebea43a364b2336
|
|
44c0ce2d4b
am: b2fd528390
Change-Id: I5ff7efe491562200ca149bda66879b52195197ae
|
|
am: 44c0ce2d4b
Change-Id: I6ad3bdd5c5c0bc836e07100ee230f144f6418c01
|
|
am: 627667649b
Change-Id: I15ad57f88c06de3064793cbc4c7b3bc09e04315a
|
|
am: ad2a78606b
Change-Id: I6a2f9de98e165fee6f8a3ea911b915393764dde6
|
|
Use the ion.h copy in system/core/libion since it's frozen.
Test: Builds.
Change-Id: Iabbd89691a9ca6b25fde320180b80aba3769b6c7
|
|
Bug: 37647680
Bug: 37648355
Test: compile, manual
Change-Id: I7e5f416f25852b59531dad929e93881533ddf5d8
|
|
Bug: 37647680
Bug: 37648355
Test: compile, manual
Change-Id: Idcf26a65cad5f6694118c68861e6d409ac187940
|
|
am: 5771ac857c
Change-Id: I6bd70326e3cb394f8fe9cb7200d19d149e877bef
|
|
am: ddb9b7fce0
Change-Id: I2a90a68a2363455e8bc4f5713accd1804c7a1fde
|
|
Bug: 37908039
BZ: 51478
Test: netflix, youtube, play movies
Regardless of the early_suspend system call.
Change-Id: Ib72396260b7532482923c33c890573dbfd40aa00
Signed-off-by: Austin Hu <austin.hu@intel.com>
(cherry picked from commit 89ee4ad98fb679a8b69c31b6034c229e05716d5c)
|
|
am: 7fe2b490b8
Change-Id: I7f14c1d607c052ca6245b99e093e525d8356a87a
|
|
am: cc0ec6239a
Change-Id: I7008770de3ac554e4ff162795b3538219268ebcf
|
|
BZ: IMINAN-51481
Fixes: 38135184
Test: Netflix, play movies, youtube
Change-Id: I67c5f1353f008035afa2492937a12818c1bff1a1
Signed-off-by: Austin Hu <austin.hu@intel.com>
(cherry picked from commit fe3bb5afb0875ca02b6232ec860ae2e33b0b4d17)
|
|
am: 4e326a4e05
Change-Id: I345be17706391e7e0a1537d812017d97a9945f25
|
|
am: b4a718c0d8
Change-Id: I0a228bf5d022169fc2cba200026a610f9e2eefe4
|
|
Bug: 33487853
Fixed by going through the normal post procedure, and creating the
corrsponding retire fence for the frame.
But actually display driver won't do real page flip by updating
the register(s) of Display Controller, to display the black screen.
Because, Anniedale Display Controller used in Fugu has a HARDWARE
bug that its pipe would hang if all of the attached RGB/YUV planes
are off.
Test: netflix, play movies, youtube
Change-Id: I9911c5e159c86fb46c02c958cb68d5c9e83856ae
Signed-off-by: Austin Hu <austin.hu@intel.com>
(cherry picked from commit eb65cfa1356880438b59009d585ebcb35ec1030f)
|
|
am: 8189283834
Change-Id: I0ee5fc85f95c330e26d8daf27a6def507a598e03
|
|
am: e7759c7536
Change-Id: I51baef38c2f4a882474bc828ff822a955148d796
|
|
BZ: IMINAN-51402
Author: Hao Liu <hao.liu@intel.com>
It fixes the regression caused by patch
https://partner-android-review.googlesource.com/#/c/759127/.
After HwcLayerList::initialize(), layers are attached with planes.
Then HwcLayerList::update() is invoked, it will DOUBLE-CHECK whether
planes are assigned properly. If not, the relevant layer's type is
set to FORCE_FRAMEBUFFER, which means it is attached to a wrong plane
or it should not be assigned to any plane at all. Then, we re-invoke
initialize() to assign planes for the 2nd time.
In that case, we should not reset the FORCE_FRAMEBUFFER layers' type.
Although the case above rarely happens, here's an example:
When we play some INTERLACED video, the real height of the video is
only half of its original label. It may trigger the planes assigned
to be invalid, then we need to withdraw the plane and do the
"FORCE_FRAMEBUFFER" stuff.
Test: netflix, playmovies, youtube
Change-Id: I730fa11ff1541a9e383778811607155c765b48e9
Signed-off-by: Austin Hu <austin.hu@intel.com>
|
|
am: ec8704c27f
Change-Id: I6522a4d418c78c5571dc9ff72f01fae26e2b0400
|
|
am: 56085c1fa6
Change-Id: Idf08a0d7575b530a256fac40a1702114cbf5799f
|
|
Test: compiled & run
Bug: 36869708
Change-Id: If83078847fb3516ae4a972e19bc788007090b585
|
|
This reverts commit be5caddb7d82f64ce765752e7451240f6a11a589.
Bug: 35427231
Test: netflix, youtube, play movies
Change-Id: Ifde1ce1fa2dcc52367c158208cb68feaa257a672
|
|
|
|
This reverts commit 7a141a8d91c694a65c0d191ef15f9c1e350263cf.
Change-Id: I2730fb82a20d5211aef85475337f0539b2910285
|
|
Bug: 33487853
BZ: IMINAN-51296
Author: Hao Liu <hao.liu@intel.com>
And fixed the screen flickering issue when switching frames with
different FRAMEBUFFER_TARGET layers, by updating the arguments
(including gralloc buffer handle) of FBT layer.
Test: Invoked youtube video
Change-Id: Ib3eeb273526e51352dbd1fc574e0993698150c9c
Signed-off-by: Austin Hu <austin.hu@intel.com>
|
|
|
|
This reverts commit a65b73f7ceada366254fb2fd8fd4b9f24dc870f4.
Change-Id: I1cb7e7874ddd49b07ad02c404e731ec390277531
|
|
|
|
Bug: 33487853
Change-Id: Ied2bc3577d8961abfdf8e6435f5fdad626480a7a
|
|
This change updates the headers imported from the Imagination
PowerVR Rogue DDK to the latest versions. The new headers are
fully backwards compatible with the older GPU DDKs.
Support for gralloc1 (v1) has also been added. This is
implemented as runtime detection of the HAL version, rather
than a compile flag, to enable the GPU DDK to be swapped out
without a full platform rebuild. The plan is that the runtime
checks will be removed in a subsequent change once the import
of DDK 1.8 has stabilized.
The headers updated by this change are used by other projects
so this change will break the build unless it is merged in
tandem with other changes.
Test: Manual, youtube video worked
Bug: 34625842
Change-Id: I700bc8845674135ff7a2b137c04a07c9945e28f7
|
|
am: 8c2ae7ccfd
Change-Id: Iba720bcb04e650545d28af884fa3013611ee0e60
|
|
am: fd1a04c8fd
Change-Id: I32f19c36b84e5056f43da18a327a56cb2cf12091
|
|
am: 9356d26309
Change-Id: I5d8d4d8915c1a3687420435a23544d12c85e2273
|
|
am: a4a351be03
Change-Id: Ib4751684de41f94fad9c583edc09fe4d77774e4b
|
|
am: 088aa2e96c
Change-Id: I79b532e510609f0f5eab2570d36509734efc9369
|
|
am: 088aa2e96c
Change-Id: I6b6dfa2064cd1850c6cf9e640d9c35fce76eb75d
|
|
41f681bad0
am: 1ef94cc544
Change-Id: I95b520f9b14206235e23f9c3e334ac98e3c15c91
|
|
am: 41f681bad0
Change-Id: I32f943bc9caaa330c2b1b96325767ee26925d379
|
|
am: 44bf9cda81
Change-Id: I3a2e8de017918d43f9c575700b0c14c958025da1
|
|
am: b5a6873157
Change-Id: Ied23369a03f2438ed49cdd0c5180f06736c1070c
|
|
this is to fix TTM dependency which back video use case
investigations have been done to prove the innocence,
1. Overlay plane is only capable for video data format
2. watch HwcLayer::attachPlane, there is no overlay plane used
3. watch TTM apis in kernel, there is no use case besides video
Change-Id: I8263cb0a2f53db1b38081fb3d1193819fad92e71
Tracked-On: https://jira01.devtools.intel.com/browse/AW-3488
Signed-off-by: wentaoli <kevin.liu@intel.com>
(cherry picked from commit 0682b313abefa21fa5155593945c4d20be378c17)
|
|
Bug: http://b/31532493
Operator precedence means that the unary ! will apply to
hwcLayer->getUsage() before applying the unary & operator with
GRALLOC_USAGE_HW_COMPOSER. This means that these checks will always fail
(because it is defined as 0x00000800U). This code has either never
executed, or just chose a fallback path to continue working.
Test: Built without errors/warnings using new compiler.
Change-Id: Ie5d67369edd801d3944be2dba9b3cc4ec289949a
|
|
780caf6cb8 am: 2816fafd20 am: cc73821479 am: c0898b42da
am: bacd482317
Change-Id: I70b2d2e2dea56446d1a7b8edc76494bd24ced9c7
|
|
780caf6cb8 am: f5b262ecb7
am: d92bf3b8c4
Change-Id: I2a9189adae341f179a9f760e9d6e8340a837b38c
|
|
780caf6cb8 am: 2816fafd20 am: cc73821479
am: c0898b42da
Change-Id: I3b1d53ff6f2b237bc6009e823e7168543687180b
|
|
780caf6cb8 am: 2816fafd20
am: cc73821479
Change-Id: I344c0f1158a9a828a675537da8b5f4de5ff806b8
|