... as a
static vendor::mapper::IMapperProvider<IMPL> provider;
may be destructed during a clean process exit (during __cxa_finalize())
but the underlying IMPL may still be in use on other threads.
Bug: b/304343434
Test: boot Cuttlefish
Change-Id: I80687494c61210d527908b5a67375950ca45c8ea
When GraphicBufferMapper::preloadHal dlopens the default impl, make sure
the gralloc module is also dlopen'ed.
Bug: 269664560
Test: android.view.cts.PixelCopyTest#testVideoProducer on grunt
Change-Id: I57418338d5212fb9a2c1ea82eba63d855f69ff32
We need to check the first api level that the vendor shipped with.
Also check P010 support in the preamble of the relevant test rather than in allocate(), since otherwise either allocate() may incorrectly skip some tests or allocate() may succeed but with the wrong layout information.
Bug: 257442231
Test: VtsHalGraphicsMapperV4_0TargetTest
Change-Id: I00ae5aa01117704158c38f0689e91df99473e208
Merged-In: I00ae5aa01117704158c38f0689e91df99473e208
We need to check the first api level that the vendor shipped with.
Also check P010 support in the preamble of the relevant test rather than in allocate(), since otherwise either allocate() may incorrectly skip some tests or allocate() may succeed but with the wrong layout information.
Bug: 257442231
Test: VtsHalGraphicsMapperV4_0TargetTest
Change-Id: I00ae5aa01117704158c38f0689e91df99473e208
sizeInBits for components in RAW formats do not have to be invalid even if the bits are not contiguous.
Bug: 265348037
Test: atest VtsHalGraphicsMapperV4_0TargetTest
Change-Id: I69f58226e988c20b8ba5499ffec4fa0e2b2a68df
Signed-off-by: Devika Krishnadas <kdevika@google.com>
* Changes the compatbility matrix version to 5.0 & marks optional
* Tweaks the interface name in case it doesn't get removed soon
(otherwise readme looks silly)
* Adds a README.md to mapper/stable-c
* Adds VTS test to validate versions match
Test: build + mapper VTS
Change-Id: I9b1256e07aaf2876d579a5487d2051c60097fb70
Specifically include the header encoding. Initially
omitted as it's a bit redundant, the value in having
bit-for-bit identical encoding between gralloc4 &
imapper5 seems worthwhile enough to keep it.
Test: impltests + VtsHalGraphicsMapperStableC
Change-Id: Iee37bb97acf40362c301a06f9118938b1a0c2cd9
Specifically include the header encoding. Initially
omitted as it's a bit redundant, the value in having
bit-for-bit identical encoding between gralloc4 &
imapper5 seems worthwhile enough to keep it.
Test: impltests + VtsHalGraphicsMapperStableC
Change-Id: Iee37bb97acf40362c301a06f9118938b1a0c2cd9
Remove all subfolder OWNERS in favor of single top-level
Update top-level OWNERS
Fixes: 230785548
Test: n/a
Change-Id: I9e4f4fa7938e31f0caf864836889f8ebb5ca5675
... by updating the toYCbCrLayout function to more closely follow
the requirements found in the comments of `android_flex_plane_t`.
Bug: b/244411455
Test: untested.
Change-Id: I1173253e25583336bf480160d4f6ac9f72882971
interfaces.
- Add new Android.bp in graphics folder and wrap composer and common
AIDL in to separate cc_defaults.
- remove composer3 dependency from allocator's VTS .bp file.
Bug: 243429120
Test: builds
Change-Id: Ia91e4ab87b7ac86248094317185b317d5604e654
We used "default" value for the allocator in the test
and this value was used as is for the AIDL allocator.
This caused the test to fail because we were not able to acquire the
appropriate allocator as AIDL allocator instance is not named "default".
Update the test to use the correct instance name for AIDL allocator,
if available.
Test: atest VtsHalGraphicsComposerV2_1TargetTest
atest VtsHalGraphicsComposerV2_2TargetTest
atest VtsHalGraphicsComposerV2_3TargetTest
atest VtsHalGraphicsComposerV2_4TargetTest
atest VtsHalGraphicsComposer3_TargetTest
BUG: 234671596
test results after updates:
VTS 2.1 : http://ab/I84600010058657636
VTS 2.2 : http://ab/I51800010058498973
VTS 2.3 : http://ab/I87700010058914863
VTS 2.4 : http://ab/I88900010058592031
VTS 3.0 : http://ab/I96200010058838905
Change-Id: I04ae1a18d757cfd941d4929ad08f6bb2c8643f76
When there is no HIDL IAllocator installed on the device, then these
tests must still run.
Bug: 231982605
Test: VtsHalGraphicsMapperV4_0TargetTest
Change-Id: If7503d398c03086df470971cc2c10029270525f9
The common decode function for optional gralloc4 metadata
`android::gralloc4::decodeOptionalMetadata()` treats an empty
input `hidl_vec<uint8_t>` as `std::nullopt` and not a bad
value. aosp/1907915 previously tried to update the encoding
and decoding of optional metadata in order to distinguish
this case but the change broke GRF. Instead, let's update the
SetBadMetadata test to use some actual incorrect data for the
optional metadata tests instead of an empty vec.
Bug: b/207388558
Test: cvd start
Test: VtsHalGraphicsMapperV4_0TargetTest
Change-Id: I87c20930cb701e87c89489bd0359cce0522cfe44
P010 is mandated for vendor versions that are at least api level 33.
Gralloc versions below this may not support P010 due to lack of
enforcement, but tests for already-shipping devices must still pass.
Bug: 220360702
Test: VtsHalGraphicsMapperV3_0TargetTest
Change-Id: I7ba268713bada133491813aa3c9bd87fdecd2fef
