| Message ID | 20241127092632.3145984-1-chenghaoyang@chromium.org |
|---|---|
| Headers | show |
| Series |
|
| Related | show |
Hi Harevy Please note that our CI system has detected 10 test failures on CTS when testing this series: https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67301704 https://yvain.ideasonboard.com/scheduler/job/84 Our baseline with CTS is 7 test failures. We know CTS failures might be a fluke but I'm wondering if you have run CTS on your systems and detected similar results or not. Thanks j On Wed, Nov 27, 2024 at 09:25:50AM +0000, Harvey Yang wrote: > Hi folks, > > This series of patches adds a new signal metadataAvailable in > libcamera::Camera, and supports partial result in Android adapter. > > Although Android allows a partial result to contain both buffers and > metadata, we probably don't need such a feature. Splitting them into > different partial results should fit both MediaTek's and Intel's > requirements. > > This patch passed gitlab pipeline: > https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/pipelines/1317276 > > There are still some discussion of the new signal in the previous patch, > with Jacopo. I'll update the patch when we reach agreement. > > I understand that the last patch is a bit large, while I don't know how > to split that furthermore. Let me know if there are some pieces that can > be standalone patches. > > BR, > Harvey > > > Han-Lin Chen (2): > libcamera: Camera: Add signals for completion of metadata as a partial > result > android: Add JpegExifMetadata to store tags setting into Exif > > Harvey Yang (7): > android: Move StreamBuffer out of Camera3RequestDescriptor > android: Set StreamBuffer::srcBuffer in > CameraDevice::processCaptureRequest > android: Migrate StreamBuffer::internalBuffer to > Camera3RequestDescriptor > android: Add CameraDevice::sendCaptureResult() > android: Cleanup CAMERA3_MSG_ERROR_REQUEST > android: Remove Camera3RequestDescriptor::streamsProcessMutex_ > android: Support partial results > > include/libcamera/camera.h | 1 + > include/libcamera/internal/pipeline_handler.h | 1 + > include/libcamera/internal/request.h | 4 + > include/libcamera/request.h | 1 + > src/android/camera_capabilities.cpp | 11 +- > src/android/camera_capabilities.h | 2 + > src/android/camera_device.cpp | 926 ++++++++++++------ > src/android/camera_device.h | 46 +- > src/android/camera_request.cpp | 94 +- > src/android/camera_request.h | 96 +- > src/android/camera_stream.cpp | 20 +- > src/android/camera_stream.h | 10 +- > src/android/jpeg/encoder.h | 2 +- > src/android/jpeg/encoder_jea.cpp | 2 +- > src/android/jpeg/encoder_jea.h | 2 +- > src/android/jpeg/encoder_libjpeg.cpp | 2 +- > src/android/jpeg/encoder_libjpeg.h | 2 +- > src/android/jpeg/post_processor_jpeg.cpp | 16 +- > src/android/jpeg/post_processor_jpeg.h | 2 +- > src/android/post_processor.h | 4 +- > src/android/yuv/post_processor_yuv.cpp | 2 +- > src/android/yuv/post_processor_yuv.h | 2 +- > src/libcamera/camera.cpp | 6 + > src/libcamera/pipeline_handler.cpp | 41 + > src/libcamera/request.cpp | 21 + > 25 files changed, 908 insertions(+), 408 deletions(-) > > -- > 2.47.0.338.g60cca15819-goog >
A second run of CTS reported 7 failures https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67365473 The three additional ones were likely flukes then. On Thu, Nov 28, 2024 at 10:22:23AM +0100, Jacopo Mondi wrote: > Hi Harevy > > Please note that our CI system has detected 10 test failures on CTS > when testing this series: > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67301704 > https://yvain.ideasonboard.com/scheduler/job/84 > > Our baseline with CTS is 7 test failures. > > We know CTS failures might be a fluke but I'm wondering if you have > run CTS on your systems and detected similar results or not. > > Thanks > j > > > On Wed, Nov 27, 2024 at 09:25:50AM +0000, Harvey Yang wrote: > > Hi folks, > > > > This series of patches adds a new signal metadataAvailable in > > libcamera::Camera, and supports partial result in Android adapter. > > > > Although Android allows a partial result to contain both buffers and > > metadata, we probably don't need such a feature. Splitting them into > > different partial results should fit both MediaTek's and Intel's > > requirements. > > > > This patch passed gitlab pipeline: > > https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/pipelines/1317276 > > > > There are still some discussion of the new signal in the previous patch, > > with Jacopo. I'll update the patch when we reach agreement. > > > > I understand that the last patch is a bit large, while I don't know how > > to split that furthermore. Let me know if there are some pieces that can > > be standalone patches. > > > > BR, > > Harvey > > > > > > Han-Lin Chen (2): > > libcamera: Camera: Add signals for completion of metadata as a partial > > result > > android: Add JpegExifMetadata to store tags setting into Exif > > > > Harvey Yang (7): > > android: Move StreamBuffer out of Camera3RequestDescriptor > > android: Set StreamBuffer::srcBuffer in > > CameraDevice::processCaptureRequest > > android: Migrate StreamBuffer::internalBuffer to > > Camera3RequestDescriptor > > android: Add CameraDevice::sendCaptureResult() > > android: Cleanup CAMERA3_MSG_ERROR_REQUEST > > android: Remove Camera3RequestDescriptor::streamsProcessMutex_ > > android: Support partial results > > > > include/libcamera/camera.h | 1 + > > include/libcamera/internal/pipeline_handler.h | 1 + > > include/libcamera/internal/request.h | 4 + > > include/libcamera/request.h | 1 + > > src/android/camera_capabilities.cpp | 11 +- > > src/android/camera_capabilities.h | 2 + > > src/android/camera_device.cpp | 926 ++++++++++++------ > > src/android/camera_device.h | 46 +- > > src/android/camera_request.cpp | 94 +- > > src/android/camera_request.h | 96 +- > > src/android/camera_stream.cpp | 20 +- > > src/android/camera_stream.h | 10 +- > > src/android/jpeg/encoder.h | 2 +- > > src/android/jpeg/encoder_jea.cpp | 2 +- > > src/android/jpeg/encoder_jea.h | 2 +- > > src/android/jpeg/encoder_libjpeg.cpp | 2 +- > > src/android/jpeg/encoder_libjpeg.h | 2 +- > > src/android/jpeg/post_processor_jpeg.cpp | 16 +- > > src/android/jpeg/post_processor_jpeg.h | 2 +- > > src/android/post_processor.h | 4 +- > > src/android/yuv/post_processor_yuv.cpp | 2 +- > > src/android/yuv/post_processor_yuv.h | 2 +- > > src/libcamera/camera.cpp | 6 + > > src/libcamera/pipeline_handler.cpp | 41 + > > src/libcamera/request.cpp | 21 + > > 25 files changed, 908 insertions(+), 408 deletions(-) > > > > -- > > 2.47.0.338.g60cca15819-goog > >
Hi Jacopo, On Thu, Nov 28, 2024 at 9:00 PM Jacopo Mondi <jacopo.mondi@ideasonboard.com> wrote: > > A second run of CTS reported 7 failures > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67365473 > > The three additional ones were likely flukes then. > > On Thu, Nov 28, 2024 at 10:22:23AM +0100, Jacopo Mondi wrote: > > Hi Harevy > > > > Please note that our CI system has detected 10 test failures on CTS > > when testing this series: > > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67301704 > > https://yvain.ideasonboard.com/scheduler/job/84 > > > > Our baseline with CTS is 7 test failures. > > > > We know CTS failures might be a fluke but I'm wondering if you have > > run CTS on your systems and detected similar results or not. Yeah v2 seems to fail more CTS tests than the previous version. In my last trial though, the tot I have on gitlab [1] passes as many CTS tests as the origin/master does. I'll run CTS again before the next version of series. BR, Harvey [1]: https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/commit/97b8c426656755b4a6e21cc8d8397ccd92395481 > > > > Thanks > > j > > > > > > On Wed, Nov 27, 2024 at 09:25:50AM +0000, Harvey Yang wrote: > > > Hi folks, > > > > > > This series of patches adds a new signal metadataAvailable in > > > libcamera::Camera, and supports partial result in Android adapter. > > > > > > Although Android allows a partial result to contain both buffers and > > > metadata, we probably don't need such a feature. Splitting them into > > > different partial results should fit both MediaTek's and Intel's > > > requirements. > > > > > > This patch passed gitlab pipeline: > > > https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/pipelines/1317276 > > > > > > There are still some discussion of the new signal in the previous patch, > > > with Jacopo. I'll update the patch when we reach agreement. > > > > > > I understand that the last patch is a bit large, while I don't know how > > > to split that furthermore. Let me know if there are some pieces that can > > > be standalone patches. > > > > > > BR, > > > Harvey > > > > > > > > > Han-Lin Chen (2): > > > libcamera: Camera: Add signals for completion of metadata as a partial > > > result > > > android: Add JpegExifMetadata to store tags setting into Exif > > > > > > Harvey Yang (7): > > > android: Move StreamBuffer out of Camera3RequestDescriptor > > > android: Set StreamBuffer::srcBuffer in > > > CameraDevice::processCaptureRequest > > > android: Migrate StreamBuffer::internalBuffer to > > > Camera3RequestDescriptor > > > android: Add CameraDevice::sendCaptureResult() > > > android: Cleanup CAMERA3_MSG_ERROR_REQUEST > > > android: Remove Camera3RequestDescriptor::streamsProcessMutex_ > > > android: Support partial results > > > > > > include/libcamera/camera.h | 1 + > > > include/libcamera/internal/pipeline_handler.h | 1 + > > > include/libcamera/internal/request.h | 4 + > > > include/libcamera/request.h | 1 + > > > src/android/camera_capabilities.cpp | 11 +- > > > src/android/camera_capabilities.h | 2 + > > > src/android/camera_device.cpp | 926 ++++++++++++------ > > > src/android/camera_device.h | 46 +- > > > src/android/camera_request.cpp | 94 +- > > > src/android/camera_request.h | 96 +- > > > src/android/camera_stream.cpp | 20 +- > > > src/android/camera_stream.h | 10 +- > > > src/android/jpeg/encoder.h | 2 +- > > > src/android/jpeg/encoder_jea.cpp | 2 +- > > > src/android/jpeg/encoder_jea.h | 2 +- > > > src/android/jpeg/encoder_libjpeg.cpp | 2 +- > > > src/android/jpeg/encoder_libjpeg.h | 2 +- > > > src/android/jpeg/post_processor_jpeg.cpp | 16 +- > > > src/android/jpeg/post_processor_jpeg.h | 2 +- > > > src/android/post_processor.h | 4 +- > > > src/android/yuv/post_processor_yuv.cpp | 2 +- > > > src/android/yuv/post_processor_yuv.h | 2 +- > > > src/libcamera/camera.cpp | 6 + > > > src/libcamera/pipeline_handler.cpp | 41 + > > > src/libcamera/request.cpp | 21 + > > > 25 files changed, 908 insertions(+), 408 deletions(-) > > > > > > -- > > > 2.47.0.338.g60cca15819-goog > > >
Hi Harvey On Wed, Dec 04, 2024 at 09:56:31PM +0800, Cheng-Hao Yang wrote: > Hi Jacopo, > > On Thu, Nov 28, 2024 at 9:00 PM Jacopo Mondi > <jacopo.mondi@ideasonboard.com> wrote: > > > > A second run of CTS reported 7 failures > > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67365473 > > > > The three additional ones were likely flukes then. > > > > On Thu, Nov 28, 2024 at 10:22:23AM +0100, Jacopo Mondi wrote: > > > Hi Harevy > > > > > > Please note that our CI system has detected 10 test failures on CTS > > > when testing this series: > > > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67301704 > > > https://yvain.ideasonboard.com/scheduler/job/84 > > > > > > Our baseline with CTS is 7 test failures. > > > > > > We know CTS failures might be a fluke but I'm wondering if you have > > > run CTS on your systems and detected similar results or not. > > Yeah v2 seems to fail more CTS tests than the previous version. > In my last trial though, the tot I have on gitlab [1] passes as many > CTS tests as the origin/master does. I see the test not being run at all https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/jobs/67693898 I think this is intended as I don't think we allow forked repositories to run CTS on our test infrastructure (I guess??) ? What CTS run are you referring to ? > > I'll run CTS again before the next version of series. > > BR, > Harvey > > [1]: https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/commit/97b8c426656755b4a6e21cc8d8397ccd92395481 > > > > > > > Thanks > > > j > > > > > > > > > On Wed, Nov 27, 2024 at 09:25:50AM +0000, Harvey Yang wrote: > > > > Hi folks, > > > > > > > > This series of patches adds a new signal metadataAvailable in > > > > libcamera::Camera, and supports partial result in Android adapter. > > > > > > > > Although Android allows a partial result to contain both buffers and > > > > metadata, we probably don't need such a feature. Splitting them into > > > > different partial results should fit both MediaTek's and Intel's > > > > requirements. > > > > > > > > This patch passed gitlab pipeline: > > > > https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/pipelines/1317276 > > > > > > > > There are still some discussion of the new signal in the previous patch, > > > > with Jacopo. I'll update the patch when we reach agreement. > > > > > > > > I understand that the last patch is a bit large, while I don't know how > > > > to split that furthermore. Let me know if there are some pieces that can > > > > be standalone patches. > > > > > > > > BR, > > > > Harvey > > > > > > > > > > > > Han-Lin Chen (2): > > > > libcamera: Camera: Add signals for completion of metadata as a partial > > > > result > > > > android: Add JpegExifMetadata to store tags setting into Exif > > > > > > > > Harvey Yang (7): > > > > android: Move StreamBuffer out of Camera3RequestDescriptor > > > > android: Set StreamBuffer::srcBuffer in > > > > CameraDevice::processCaptureRequest > > > > android: Migrate StreamBuffer::internalBuffer to > > > > Camera3RequestDescriptor > > > > android: Add CameraDevice::sendCaptureResult() > > > > android: Cleanup CAMERA3_MSG_ERROR_REQUEST > > > > android: Remove Camera3RequestDescriptor::streamsProcessMutex_ > > > > android: Support partial results > > > > > > > > include/libcamera/camera.h | 1 + > > > > include/libcamera/internal/pipeline_handler.h | 1 + > > > > include/libcamera/internal/request.h | 4 + > > > > include/libcamera/request.h | 1 + > > > > src/android/camera_capabilities.cpp | 11 +- > > > > src/android/camera_capabilities.h | 2 + > > > > src/android/camera_device.cpp | 926 ++++++++++++------ > > > > src/android/camera_device.h | 46 +- > > > > src/android/camera_request.cpp | 94 +- > > > > src/android/camera_request.h | 96 +- > > > > src/android/camera_stream.cpp | 20 +- > > > > src/android/camera_stream.h | 10 +- > > > > src/android/jpeg/encoder.h | 2 +- > > > > src/android/jpeg/encoder_jea.cpp | 2 +- > > > > src/android/jpeg/encoder_jea.h | 2 +- > > > > src/android/jpeg/encoder_libjpeg.cpp | 2 +- > > > > src/android/jpeg/encoder_libjpeg.h | 2 +- > > > > src/android/jpeg/post_processor_jpeg.cpp | 16 +- > > > > src/android/jpeg/post_processor_jpeg.h | 2 +- > > > > src/android/post_processor.h | 4 +- > > > > src/android/yuv/post_processor_yuv.cpp | 2 +- > > > > src/android/yuv/post_processor_yuv.h | 2 +- > > > > src/libcamera/camera.cpp | 6 + > > > > src/libcamera/pipeline_handler.cpp | 41 + > > > > src/libcamera/request.cpp | 21 + > > > > 25 files changed, 908 insertions(+), 408 deletions(-) > > > > > > > > -- > > > > 2.47.0.338.g60cca15819-goog > > > >
Hi Jacopo, On Wed, Dec 4, 2024 at 10:03 PM Jacopo Mondi <jacopo.mondi@ideasonboard.com> wrote: > > Hi Harvey > > On Wed, Dec 04, 2024 at 09:56:31PM +0800, Cheng-Hao Yang wrote: > > Hi Jacopo, > > > > On Thu, Nov 28, 2024 at 9:00 PM Jacopo Mondi > > <jacopo.mondi@ideasonboard.com> wrote: > > > > > > A second run of CTS reported 7 failures > > > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67365473 > > > > > > The three additional ones were likely flukes then. > > > > > > On Thu, Nov 28, 2024 at 10:22:23AM +0100, Jacopo Mondi wrote: > > > > Hi Harevy > > > > > > > > Please note that our CI system has detected 10 test failures on CTS > > > > when testing this series: > > > > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67301704 > > > > https://yvain.ideasonboard.com/scheduler/job/84 > > > > > > > > Our baseline with CTS is 7 test failures. > > > > > > > > We know CTS failures might be a fluke but I'm wondering if you have > > > > run CTS on your systems and detected similar results or not. > > > > Yeah v2 seems to fail more CTS tests than the previous version. > > In my last trial though, the tot I have on gitlab [1] passes as many > > CTS tests as the origin/master does. > > I see the test not being run at all > https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/jobs/67693898 > > I think this is intended as I don't think we allow forked repositories > to run CTS on our test infrastructure (I guess??) ? > > What CTS run are you referring to ? I ran CTS (android-cts-13_r7) manually with my own soraka DUT. We can also use the result on gitlab instead. Do I need to ask for your help to run it everytime? BR, Harvey > > > > > I'll run CTS again before the next version of series. > > > > BR, > > Harvey > > > > [1]: https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/commit/97b8c426656755b4a6e21cc8d8397ccd92395481 > > > > > > > > > > Thanks > > > > j > > > > > > > > > > > > On Wed, Nov 27, 2024 at 09:25:50AM +0000, Harvey Yang wrote: > > > > > Hi folks, > > > > > > > > > > This series of patches adds a new signal metadataAvailable in > > > > > libcamera::Camera, and supports partial result in Android adapter. > > > > > > > > > > Although Android allows a partial result to contain both buffers and > > > > > metadata, we probably don't need such a feature. Splitting them into > > > > > different partial results should fit both MediaTek's and Intel's > > > > > requirements. > > > > > > > > > > This patch passed gitlab pipeline: > > > > > https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/pipelines/1317276 > > > > > > > > > > There are still some discussion of the new signal in the previous patch, > > > > > with Jacopo. I'll update the patch when we reach agreement. > > > > > > > > > > I understand that the last patch is a bit large, while I don't know how > > > > > to split that furthermore. Let me know if there are some pieces that can > > > > > be standalone patches. > > > > > > > > > > BR, > > > > > Harvey > > > > > > > > > > > > > > > Han-Lin Chen (2): > > > > > libcamera: Camera: Add signals for completion of metadata as a partial > > > > > result > > > > > android: Add JpegExifMetadata to store tags setting into Exif > > > > > > > > > > Harvey Yang (7): > > > > > android: Move StreamBuffer out of Camera3RequestDescriptor > > > > > android: Set StreamBuffer::srcBuffer in > > > > > CameraDevice::processCaptureRequest > > > > > android: Migrate StreamBuffer::internalBuffer to > > > > > Camera3RequestDescriptor > > > > > android: Add CameraDevice::sendCaptureResult() > > > > > android: Cleanup CAMERA3_MSG_ERROR_REQUEST > > > > > android: Remove Camera3RequestDescriptor::streamsProcessMutex_ > > > > > android: Support partial results > > > > > > > > > > include/libcamera/camera.h | 1 + > > > > > include/libcamera/internal/pipeline_handler.h | 1 + > > > > > include/libcamera/internal/request.h | 4 + > > > > > include/libcamera/request.h | 1 + > > > > > src/android/camera_capabilities.cpp | 11 +- > > > > > src/android/camera_capabilities.h | 2 + > > > > > src/android/camera_device.cpp | 926 ++++++++++++------ > > > > > src/android/camera_device.h | 46 +- > > > > > src/android/camera_request.cpp | 94 +- > > > > > src/android/camera_request.h | 96 +- > > > > > src/android/camera_stream.cpp | 20 +- > > > > > src/android/camera_stream.h | 10 +- > > > > > src/android/jpeg/encoder.h | 2 +- > > > > > src/android/jpeg/encoder_jea.cpp | 2 +- > > > > > src/android/jpeg/encoder_jea.h | 2 +- > > > > > src/android/jpeg/encoder_libjpeg.cpp | 2 +- > > > > > src/android/jpeg/encoder_libjpeg.h | 2 +- > > > > > src/android/jpeg/post_processor_jpeg.cpp | 16 +- > > > > > src/android/jpeg/post_processor_jpeg.h | 2 +- > > > > > src/android/post_processor.h | 4 +- > > > > > src/android/yuv/post_processor_yuv.cpp | 2 +- > > > > > src/android/yuv/post_processor_yuv.h | 2 +- > > > > > src/libcamera/camera.cpp | 6 + > > > > > src/libcamera/pipeline_handler.cpp | 41 + > > > > > src/libcamera/request.cpp | 21 + > > > > > 25 files changed, 908 insertions(+), 408 deletions(-) > > > > > > > > > > -- > > > > > 2.47.0.338.g60cca15819-goog > > > > >
Hi Harvey On Wed, Dec 04, 2024 at 10:10:28PM +0800, Cheng-Hao Yang wrote: > Hi Jacopo, > > On Wed, Dec 4, 2024 at 10:03 PM Jacopo Mondi > <jacopo.mondi@ideasonboard.com> wrote: > > > > Hi Harvey > > > > On Wed, Dec 04, 2024 at 09:56:31PM +0800, Cheng-Hao Yang wrote: > > > Hi Jacopo, > > > > > > On Thu, Nov 28, 2024 at 9:00 PM Jacopo Mondi > > > <jacopo.mondi@ideasonboard.com> wrote: > > > > > > > > A second run of CTS reported 7 failures > > > > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67365473 > > > > > > > > The three additional ones were likely flukes then. > > > > > > > > On Thu, Nov 28, 2024 at 10:22:23AM +0100, Jacopo Mondi wrote: > > > > > Hi Harevy > > > > > > > > > > Please note that our CI system has detected 10 test failures on CTS > > > > > when testing this series: > > > > > https://gitlab.freedesktop.org/camera/libcamera/-/jobs/67301704 > > > > > https://yvain.ideasonboard.com/scheduler/job/84 > > > > > > > > > > Our baseline with CTS is 7 test failures. > > > > > > > > > > We know CTS failures might be a fluke but I'm wondering if you have > > > > > run CTS on your systems and detected similar results or not. > > > > > > Yeah v2 seems to fail more CTS tests than the previous version. > > > In my last trial though, the tot I have on gitlab [1] passes as many > > > CTS tests as the origin/master does. > > > > I see the test not being run at all > > https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/jobs/67693898 > > > > I think this is intended as I don't think we allow forked repositories > > to run CTS on our test infrastructure (I guess??) ? > > > > What CTS run are you referring to ? > > I ran CTS (android-cts-13_r7) manually with my own soraka DUT. Ah sorry, I didn't get it > We can also use the result on gitlab instead. Do I need to ask for > your help to run it everytime? > I don't think we can (and should) allow forks of camera/libcamera to run tests on our physical infrastructure, only series collected from patchwork that touch src/android are run on our LAVA lab. I'll ask if it is possible or desirable to make singular execptions. > BR, > Harvey > > > > > > > > > I'll run CTS again before the next version of series. > > > > > > BR, > > > Harvey > > > > > > [1]: https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/commit/97b8c426656755b4a6e21cc8d8397ccd92395481 > > > > > > > > > > > > > Thanks > > > > > j > > > > > > > > > > > > > > > On Wed, Nov 27, 2024 at 09:25:50AM +0000, Harvey Yang wrote: > > > > > > Hi folks, > > > > > > > > > > > > This series of patches adds a new signal metadataAvailable in > > > > > > libcamera::Camera, and supports partial result in Android adapter. > > > > > > > > > > > > Although Android allows a partial result to contain both buffers and > > > > > > metadata, we probably don't need such a feature. Splitting them into > > > > > > different partial results should fit both MediaTek's and Intel's > > > > > > requirements. > > > > > > > > > > > > This patch passed gitlab pipeline: > > > > > > https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/pipelines/1317276 > > > > > > > > > > > > There are still some discussion of the new signal in the previous patch, > > > > > > with Jacopo. I'll update the patch when we reach agreement. > > > > > > > > > > > > I understand that the last patch is a bit large, while I don't know how > > > > > > to split that furthermore. Let me know if there are some pieces that can > > > > > > be standalone patches. > > > > > > > > > > > > BR, > > > > > > Harvey > > > > > > > > > > > > > > > > > > Han-Lin Chen (2): > > > > > > libcamera: Camera: Add signals for completion of metadata as a partial > > > > > > result > > > > > > android: Add JpegExifMetadata to store tags setting into Exif > > > > > > > > > > > > Harvey Yang (7): > > > > > > android: Move StreamBuffer out of Camera3RequestDescriptor > > > > > > android: Set StreamBuffer::srcBuffer in > > > > > > CameraDevice::processCaptureRequest > > > > > > android: Migrate StreamBuffer::internalBuffer to > > > > > > Camera3RequestDescriptor > > > > > > android: Add CameraDevice::sendCaptureResult() > > > > > > android: Cleanup CAMERA3_MSG_ERROR_REQUEST > > > > > > android: Remove Camera3RequestDescriptor::streamsProcessMutex_ > > > > > > android: Support partial results > > > > > > > > > > > > include/libcamera/camera.h | 1 + > > > > > > include/libcamera/internal/pipeline_handler.h | 1 + > > > > > > include/libcamera/internal/request.h | 4 + > > > > > > include/libcamera/request.h | 1 + > > > > > > src/android/camera_capabilities.cpp | 11 +- > > > > > > src/android/camera_capabilities.h | 2 + > > > > > > src/android/camera_device.cpp | 926 ++++++++++++------ > > > > > > src/android/camera_device.h | 46 +- > > > > > > src/android/camera_request.cpp | 94 +- > > > > > > src/android/camera_request.h | 96 +- > > > > > > src/android/camera_stream.cpp | 20 +- > > > > > > src/android/camera_stream.h | 10 +- > > > > > > src/android/jpeg/encoder.h | 2 +- > > > > > > src/android/jpeg/encoder_jea.cpp | 2 +- > > > > > > src/android/jpeg/encoder_jea.h | 2 +- > > > > > > src/android/jpeg/encoder_libjpeg.cpp | 2 +- > > > > > > src/android/jpeg/encoder_libjpeg.h | 2 +- > > > > > > src/android/jpeg/post_processor_jpeg.cpp | 16 +- > > > > > > src/android/jpeg/post_processor_jpeg.h | 2 +- > > > > > > src/android/post_processor.h | 4 +- > > > > > > src/android/yuv/post_processor_yuv.cpp | 2 +- > > > > > > src/android/yuv/post_processor_yuv.h | 2 +- > > > > > > src/libcamera/camera.cpp | 6 + > > > > > > src/libcamera/pipeline_handler.cpp | 41 + > > > > > > src/libcamera/request.cpp | 21 + > > > > > > 25 files changed, 908 insertions(+), 408 deletions(-) > > > > > > > > > > > > -- > > > > > > 2.47.0.338.g60cca15819-goog > > > > > >
Hi folks, This series of patches adds a new signal metadataAvailable in libcamera::Camera, and supports partial result in Android adapter. Although Android allows a partial result to contain both buffers and metadata, we probably don't need such a feature. Splitting them into different partial results should fit both MediaTek's and Intel's requirements. This patch passed gitlab pipeline: https://gitlab.freedesktop.org/chenghaoyang/libcamera/-/pipelines/1317276 There are still some discussion of the new signal in the previous patch, with Jacopo. I'll update the patch when we reach agreement. I understand that the last patch is a bit large, while I don't know how to split that furthermore. Let me know if there are some pieces that can be standalone patches. BR, Harvey Han-Lin Chen (2): libcamera: Camera: Add signals for completion of metadata as a partial result android: Add JpegExifMetadata to store tags setting into Exif Harvey Yang (7): android: Move StreamBuffer out of Camera3RequestDescriptor android: Set StreamBuffer::srcBuffer in CameraDevice::processCaptureRequest android: Migrate StreamBuffer::internalBuffer to Camera3RequestDescriptor android: Add CameraDevice::sendCaptureResult() android: Cleanup CAMERA3_MSG_ERROR_REQUEST android: Remove Camera3RequestDescriptor::streamsProcessMutex_ android: Support partial results include/libcamera/camera.h | 1 + include/libcamera/internal/pipeline_handler.h | 1 + include/libcamera/internal/request.h | 4 + include/libcamera/request.h | 1 + src/android/camera_capabilities.cpp | 11 +- src/android/camera_capabilities.h | 2 + src/android/camera_device.cpp | 926 ++++++++++++------ src/android/camera_device.h | 46 +- src/android/camera_request.cpp | 94 +- src/android/camera_request.h | 96 +- src/android/camera_stream.cpp | 20 +- src/android/camera_stream.h | 10 +- src/android/jpeg/encoder.h | 2 +- src/android/jpeg/encoder_jea.cpp | 2 +- src/android/jpeg/encoder_jea.h | 2 +- src/android/jpeg/encoder_libjpeg.cpp | 2 +- src/android/jpeg/encoder_libjpeg.h | 2 +- src/android/jpeg/post_processor_jpeg.cpp | 16 +- src/android/jpeg/post_processor_jpeg.h | 2 +- src/android/post_processor.h | 4 +- src/android/yuv/post_processor_yuv.cpp | 2 +- src/android/yuv/post_processor_yuv.h | 2 +- src/libcamera/camera.cpp | 6 + src/libcamera/pipeline_handler.cpp | 41 + src/libcamera/request.cpp | 21 + 25 files changed, 908 insertions(+), 408 deletions(-)