So camera continue to work when camera service switch to use
the @3.5 session.
Test: manually check configureStream works
Bug: 109829698
Change-Id: I4c7edcf59e68a2b4f04fee8229ed5d3c9b261c9f
Also includes:
- Add test for getPhysicalCameraCharacteristics.
- Doc update for the new HIDL API.
- Tighten boundary check for camera ID in CameraModule.
Test: vts-tradefed run commandAndExit vts --skip-all-system-status-check
--skip-preconditions --module VtsHalCameraProviderV2_4Target -l INFO
Bug: 79523700
Bug: 115969176
Bug: 116512585
Change-Id: I051d1b0c91834781a1f8d893ed5ebfa579b03774
- Add version 3.5 for ICameraDevice for physical camera characteristics
query.
- Add version 3.5 for ICameraDeviceSession to work around HIDL
versioning bug.
Test: Camera CTS
Bug: 79523700
Change-Id: I8df6cdd4ee6ac5755758510c0dc1ea1cec31aa73
Claim torch API support though there is no device
actually has a flash unit. This is to workaround
an issue in camera service where it falls back
to legacy camera HAL1 control if the device does
not report torch API support (and external camera
HAL does not implement HAL1)
Test: CtsVerifier FlashLight test
Bug: 113336515
Change-Id: I945b35fecfed3b2698b1e9bb1d8faad8699258d2
HW video encoder usually has limitation to encode very small
stream. Camera should avoid to return such streams to avoid
video encoding failure.
BUG: 110654511
Test: CTS android.hardware.camera2.cts.RecordingTest#testSupportedVideoSizes
Change-Id: I6104f2d4efcf7831d1dc6d305c82309e2fd1999d
This option allows HAL to opt in for the behavior that will free
cached buffers earlier if the libhardware HAL implementation
doesn't cache/reference the cached buffers internally.
Test: buffers are freed earlier when the property is set
Bug: 111850884
Change-Id: I3a10b288c7160c86dc7d3a30d04b5c4903917731
HW video encoder usually has limitation to encode very small
stream. Camera should avoid to return such streams to avoid
video encoding failure.
BUG: 110654511
Test: CTS android.hardware.camera2.cts.RecordingTest#testSupportedVideoSizes
Change-Id: I6104f2d4efcf7831d1dc6d305c82309e2fd1999d
If capture result metadata is to be shrinked, reserve enough space so
that CameraMetadata isn't re-allocated when std::vector expands.
If vector expansion happens, CameraMetadata is re-allocated, causing
camera_metadata_t being re-allocated as well.
Test: Camera CTS
Bug: 109672123
Change-Id: Idaf5bf409bdbbccf617614edc96c0478260b4c30
When camera HAL sends redundant metadata (reserved space >> actual
used space), create a compact copy to save IPC overhead.
Test: Camera CTS/camera app smoke test
Bug: 79838637
Change-Id: I02657deb377c43c65d111d1fc894eb3c20ad3b8f
Camera device session 3.4 has a separate result batcher
instance. The partial result count needs to get initialized
in the same way as ver. 3.2 otherwise the frame batches can
get retired prematurely.
Bug: 79106874
Test: Camera CTS,
run vts --skip-all-system-status-check --skip-preconditions
--primary-abi-only --module VtsHalCameraProviderV2_4Target -l INFO
Change-Id: I36bfb5bc448154f80f4cb94edcfd119048361099
It's generally a bad thing to have the result exploded to exceed
the shared message max depth (1MB). Print the size when such case
occurs for better debugging output.
Bug: 77665564
Change-Id: I971fca202aa3ca0fa1863e9c658659d0775a1ad4
V4L2 buffer size should fall within a reasonable range
like (0. width*height*2]. Check and return appropriate
status in case the size is not within this range.
Bug: 72261912
Test: Camera CTS
Change-Id: I1945dc734afb552ecb8d4ef74c9198eca8ce00e7
Make sure that num_physcam_metadata, physcam_ids, and physcam_metadata
returned from process_capture_result is sane.
Test: LogicalCameraDeviceTest
Bug: 77218595
Change-Id: Ic4d9823569fb86b20ab2ab21edfc32a0e9f5c9c7
When calling update-makefiles.sh there are
some unrelated changes that were missed in
previous commits.
Bug: 8675309
Test: compilation
Change-Id: I5bf67fbcc809de36bde1869ada7b835566a5198b
1. Add EXIF MAKE/MODEL tag
2. Add retry loop for V4L2 open failure
3. Do not add external camera if ExternalCameraDevice cannot be
initialized
Bug: 72261912
Bug: 72569850
Change-Id: I06df1fbbb4afabea1a9a74aca9e288b24966cb0b
If HAL calls process_capture_result with invalid buffers, the HIDL
wrapper shouldn't forward it to camera service, since the CaptureResult
structure may not be completely initialized.
Test: Run GoogleCamera
Bug: 74433802
Change-Id: Iaf20b542af5b11ffbedf709e99137b69d77a9ad3
Also update default external camera config to prevent
over promising peformance on lower end devices.
Test: API2 CTS tests except RecordingTest
Bug: 72261744
Change-Id: I20310606520099abf0f5173690f4aaf56a3e52a9
1. Update the FPS range to list (0.5*fps, fps) only
as webcams tends to skip a lot of frames and not
able to output at stable framerate.
2. Exif: don't expect focal length to present
3. Thumbnail: allow 0x0 size for no thumbnail output
4. Allow retry some ioctl during configureStream as
some webcams seems having problem in quick close
reopen operation.
Test: CTS CameraTest
Bug: 72261912
Change-Id: Ic23b7fb293b7579694c59240e854d750c842886d
V4L2 doesn't seem to allow changing fps at runtime, so we need to
wait until pipeline is idle and reconfigure V4L2 streams.
Also update the fps filtering logic to make sure supported fps
are consistent as long as the major dimension is the same.
Ex: for 4:3 webcams, all 1280*N sizes has the same fps set, or
for 16:9 webcams, all N*720 sizes has the same fps set.
Bug: 72261912
Change-Id: I534f09bebadb2f532a030a841633a8be7d3a87cc
Simplify the API for the P release
- Remove new templates
- Update docs
- Also update some metadata docs from prior changes
Test: Revised Camera CTS passes
Bug: 63629224
Change-Id: I47fb170052cf6062955278ace624a91e8a0078b3
Empty individual physical settings should be allowed after
one initial successful request.
Bug: 72524845
Test: run vts --skip-all-system-status-check --skip-preconditions
--primary-abi-only --module VtsHalCameraProviderV2_4Target -l INFO
Change-Id: I18d22a56f5f0e6e1d81a6e8c844697c02bc343a1
Release fence close was handled by freeReleaseFences so
hidl_handle should not own the native handle.
Test: running external camera on marlin
Bug: 64874137
Change-Id: I68e76d31985380eb3c502276df07d3648f67b6f1
Individual physical camera requests should only be allowed
in case the respective physical Ids are among the last
configured Hal streams.
Bug: 72524603
Test: run vts --skip-all-system-status-check --skip-preconditions
--primary-abi-only --module VtsHalCameraProviderV2_4Target -l INFO
Change-Id: Ia5715398f74f03528d7fe2e52e5b9e32ad654a2c
So that HALs without device knowledge knows the maximal size
it can lock using graphics mapper API.
Bug: 64874137 72261675
Change-Id: I7363f5c63d955ab412f9570bf8e5be2a942cf381
This code was exiting because of unsupported RequestTemplate::*
enum values that are now just ignored.
Change-Id: I37f79cbbfe99664ec173458721d71947f0d002ff
Also switch to new EXTERNAL hardware level.
Test: ImageReaderTest#testAllOutputYUVResolutions pass
Bug: 72261912
Change-Id: I95f05a20827b413bfa480b0336fff395159678eb
- When physical stream of a logical multi-camera is requested, HAL needs
to generate metadata for the physical cameras.
- In case no physical stream is requested for the logical multi-camera, no
capture result metadata is required for physical camera.
- Batch physical and logical metadata within one capture_result call.
Test: testLogicalCamera CTS test
Bug: 64691172
Change-Id: Id040620f3f0c350711d49341ab31ab88ecd94888
- Extend RequestTemplate with two new values
- Add legacy wrapper support for the new templates
Test: Builds, camera continues to run. CTS coverage of templates
tests new templates as expected.
Bug: 63629224
Change-Id: I2a44d86a815690798fa4b8327794c64716d134ee
Multi-camera devices should be able to accept and process
individual settings for all physical sensors backing the
logical camera. 'CaptureRequest' must be extended to hold
these additional physical camera settings.
Test: Manual using camera application,
camera_client_test,
run vts --skip-all-system-status-check --skip-preconditions
--primary-abi-only --module VtsHalCameraProviderV2_4Target -l INFO
Bug: 64691172
Change-Id: Ia38d1e7681a9385be7578c11e40f4e35e9101d75
- Add physical cameraId field in stream structure.
- Override processCaptureRequest due to physicalCameraId change.
- Update 3.3 metadata with logical camera characteristics.
Test: Camera CTS on Taimen, LogicalCamera CTS test on C1
Bug: 64691172
Change-Id: I65bd1ae38381ecb89fae439ae14b813c9bcc3248
According to the camera_module_t::get_number_of_cameras() description
in camera_common.h, it should only return the number of "fixed"
cameras, i.e. cameras, facing to the front and to the back. Any
cameras, with the facing value of "external," i.e. externally connected
cameras, should not be reported by it, instead they should later be
reported, using the
camera_module_callbacks_t::camera_device_status_change() callback.
However, this doesn't work. When a camera is reported this way, it is
ignored as unknown. Fix this by adding a new camera, when its status
change is reported.
Change-Id: I18874ed005f477e32fb723c00c16b67a81cde6ce
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@intel.com>
Camera devices supporting version 3.4 will be able to receive
session parameters during the stream configuration phase.
Bug: 64450664
Test: Camera CTS
run commandAndExit vts --skip-all-system-status-check
--skip-preconditions --primary-abi-only --module
VtsHalCameraProviderV2_4Target -l INFO
Change-Id: Ifd83bfe0e512fe75b63602b4aba98f4cc1cdeb53
