Integrating Hardware-accelerated Video Decoding with the Display Stack
In February 2018, Bootlin launched a crowdfunding campaign to fund thedevelopment of upstream Linux kernel support for the hardware-acceleratedvideo decoder (VPU) found on Allwinner platforms. This wo …
| Talk Title | Integrating Hardware-accelerated Video Decoding with the Display Stack |
| Speakers | Paul Kocialkowski (Embedded Linux engineer, Bootlin) |
| Conference | Open Source Summit + ELC Europe |
| Conf Tag | |
| Location | Lyon, France |
| Date | Oct 27-Nov 1, 2019 |
| URL | Talk Page |
| Slides | Talk Slides |
| Video | |
In February 2018, Bootlin launched a crowdfunding campaign to fund thedevelopment of upstream Linux kernel support for the hardware-acceleratedvideo decoder (VPU) found on Allwinner platforms. This work is part ofBootlin’s larger ongoing effort to provide upstream Linux support forthese platforms.Adding support for the VPU itself was a significant effort, involving awhole new V4L2 API (the Request API) used on top of the existing M2M API,codec-specific uAPI bits, a new driver (cedrus) as well as userspacecomponents such as a VAAPI backend and a test utility. However, the mostpainful point turned out to be the integration of the decoded frames withthe various possible display pipelines.This talk will introduce some context about hardware video decoding,the work we carried out and the major issue we encountered with displayintegration as well as lessons learned from the experience.
