On Wed, Dec 12, 2018 at 09:23:30PM +0000, Edward Cree wrote: > On 12/12/18 20:13, Alexei Starovoitov wrote: > > that won't work either. > > "bpf feature set" is a lot more than number of program > > and map types the kernel supports. There are all sorts of > > helper combinations, hooks, and verifier improvements. > > test_verifier.c must test all that. > > I don't think there is a way to make usptream test_verfier.c > > not to report failure on older kernels. > But it's not just older kernels; AIUI there are config options > that also affect this. Are you saying that test_verifier > should only be expected to run / pass on allyesconfig kernels?
not allyesconfig, but allbpfyesconfig. > I think that for the cases where we _can_ do it easily (which > seems to be precisely things like prog_type which don't require > any additional annotation of test cases) we should skip tests > that aren't supported by the running kernel. Submitters of bpf patches need to enable all bpf features in their kernels and make sure that all of the selftests/bpf are still passing. "but I don't use XDP today" or "LLVM is hard to install" is not an excuse. Before I apply any patch I build and test manually. If anything breaks means folks didn't do enough testing and karma suffers.
