On Wed, Apr 3, 2024 at 11:59 AM Brennan Vincent <brennan.vincent@xxxxxxxxxxxxxxxx> wrote: > > I am not sure what, if anything, should be done about this in the kernel, > as it seems there's a fundamental tradeoff between precision of the > verifier and performance in terms of instructions processed. To avoid such surprises in the future it's best to contribute your bpf programs as selftests. Then future verifier changes will be tested against them.
