A Push Towards Firmware-less Video Decoding By Linux Kernel Media Drivers
Veteran Linux multimedia developer Paul Kocialkowski summed up the current situation this week of many hardware platforms having a general purpose micro-controller running a non-free firmware blob for coordinating the video decoding work. It makes it easier to program with this firmware-based approach but makes the driver less free and now with recent Linux infrastructure improvements could better support dealing with the video hardware itself.
While the firmware-based video decoding makes the driver work easier, it's contingent upon the binary firmware blobs and the micro-controller running it doesn't necessarily be wasting energy on that task. With recent work on the Linux kernel's media interface, the kernel can now better support interfacing with hardware decoders directly.
Kocialkowski notes that with tapping the hardware decoder directly, it would work better for fully-free software setups, easier to update with having no firmware blob in the process, possible latency/performance advantages, and freeing up the microcontroller to do other work.
He's hoping the Linux media subsystem developers will be able to push vendors for more direct hardware support now that the kernel infrastructure can better handle it than the firmware-based video decoding. Of course, not many vendors like publicly documenting their registers and other hardware details, but we'll see where this effort leads. More details in this kernel mailing list thread.
While the firmware-based video decoding makes the driver work easier, it's contingent upon the binary firmware blobs and the micro-controller running it doesn't necessarily be wasting energy on that task. With recent work on the Linux kernel's media interface, the kernel can now better support interfacing with hardware decoders directly.
Kocialkowski notes that with tapping the hardware decoder directly, it would work better for fully-free software setups, easier to update with having no firmware blob in the process, possible latency/performance advantages, and freeing up the microcontroller to do other work.
He's hoping the Linux media subsystem developers will be able to push vendors for more direct hardware support now that the kernel infrastructure can better handle it than the firmware-based video decoding. Of course, not many vendors like publicly documenting their registers and other hardware details, but we'll see where this effort leads. More details in this kernel mailing list thread.
6 Comments