BPF: Scaling the verifier
Hi,
I'd like to attend the LSF/MM Summit 2019, the BPF track. I'm working on BPF, and the initial author of bpftool and networking BPF offload.
I'd like to discuss ways of scaling the verifier to 1M instructions, something that Alexei often mentions as a design goal :)
I'd like to share results of verifier performance analysis, and evaluation of a few possible optimization ideas. Related to that - we got rid of the last piece of shared verifier state in Oct 2017, but the global verifier lock is still there, we should probably sketch a way forward on that front.
If there is space at the back of the room, I'd also like to attend the MM discussions (in HMM area, as it relates to BPF offload and sharing BPF state with next-gen networking HW), and NVMe discussion (again, in networking context).