LinuxMusicians

https://en.wikipedia.org/wiki/Berkeley_Packet_Filter - with information on the basics of eBPF

"In his linux.conf.au 2017 talk on the eBPF in-kernel virtual machine, Brendan Gregg proclaimed that "super powers have finally come to Linux". Getting eBPF to that point has been a long road of evolution and design. While eBPF was originally used for network packet filtering, it turns out that running user-space code inside a sanity-checking virtual machine is a powerful tool for kernel developers and production engineers.
-- LWN.net: A thorough introduction to eBPF

The big performance advantage is that context switches are reduced.

More recently, the use of eBPF has been expanded. A couple of insightful videos;

How to Make Linux Microservice-Aware with Cilium and eBPF - first half is a good eBPF overview, the second half relates to security, specifically in-kernel application level firewall rules.

LPC2018 - When eBPF meets FUSE: Improving Performance of User File Systems - user space filesystems in the kernel

So my wonder is this; could eBPF be used for in-kernel audio DSP?

See also https://github.com/iovisor/bcc and https://github.com/zoidbergwill/awesome-ebpf

I think you can use kernel_fpu_begin/kernel_fpu_end around dsp code in a kernel thread since at least kernel 2.6, even directly in the interrupt thread of your sound device.

Not sure if that's a good idea:

- It's just-in-time-compiled code, so it's hardly slower but in no case faster than compiled code in user space. No win here.

- Communication with code in user space (DAW) would not be obsolete, so it's basically similar to today's architecture.

- It can run code in user space but it's designed to run in kernel space: one more place where user supplied code could possibly destabilize the whole machine's operation.One more threat here.

Sure,it's a new toy. But will it enable anything usable? Or will it just blow up the kernel a bit more?