V4L2 M2M as the Driver Framework for Video Processing IP
Talkat Embedded Linux Conference @ Open Source Summit North America 2022 · on Thursday, 2022-06-23· at 14:05· Room 203/204 (Level 2), JW Marriott Austin, 110 E 2nd Street Austin, TX 78701·
summary
Explains why V4L2 M2M framework is the right kernel driver model for memory-to-memory video IP and shows a working virtual M2M scaler built with Kimu, running on Linux kernel 5.15 and using the open source SDB image resize library. Driver implements VB2, M2M core, threaded IRQs, regmap, media controller and supports multicontext; a libcamera C++ test app queues buffers, verifies pixel-for-pixel output for a 640×480→320×240 downscale, and reports successful round trips.
AI-written summary · about
chapters (11)
- 0:00Start
- 0:32What Is V4 L2 M2m
- 2:34Typical Referral to Application Workflow
- 9:42The Device Driver
- 18:56Tri Format
- 19:49Device Run
- 22:33Private Data Structures
- 23:56Constructor
- 26:22Set Up the Formats
- 26:57Set Up the Callbacks
- 29:51Callbacks
related talks
36mCamera/ISP Drivers Using V4L2 Media Controller Framework
41mHow V4L2 Transformed To Support Embedded Cameras
41mLearn How to Support Your SoC and ISP in Libcamera
25mLibcamera: Making Complex Cameras Easy
38mVideo4Zephyr: From Basic APIs Towards a Full-fledged Video Subsystem
29mComposite Video Device Abstraction for Libcamera
42mSoftware ISP FOSS Support for MIPI Cameras- 38mVideo4Zephyr: From Basic APIs Towards a Full-fledged Video Subsystem