On 12/09, Jakub Kicinski wrote: > On Thu, 6 Dec 2018 19:39:06 -0800, Stanislav Fomichev wrote: > > Include "autoconf.h" header in the test_maps.c and selectively > > disable test_sockmap if CONFIG_BPF_STREAM_PARSER is not specified > > in the kernel config. > > When building out of tree/without autoconf.h, fall back to 'enabled'. > > > > Signed-off-by: Stanislav Fomichev <s...@google.com> > > No opinion on the merits, but if we are to skip tests I think its > better to use SKIP result. IMHO its better for person running the test > to know they are not running all the tests than have them silently > disappear. SGTM, I'll add some indication that there were X tests skipped.
> If we were to conditionally skip tests should we also tackle the > test_verifier cases which require sockmap? Yes, I was going to follow up on that, I just wanted to get some initial response on the idea of skipping the tests based on .config > #274/p prevent map lookup in sockmap Failed to create hash map 'Invalid > argument'! > FAIL > Unexpected error message! > EXP: cannot pass map_type 15 into func bpf_map_lookup_elem > RES: fd -1 is not pointing to valid bpf_map > > fd -1 is not pointing to valid bpf_map > #275/p prevent map lookup in sockhash Failed to create hash map 'Invalid > argument'! > FAIL > Unexpected error message! > EXP: cannot pass map_type 18 into func bpf_map_lookup_elem > RES: fd -1 is not pointing to valid bpf_map > > fd -1 is not pointing to valid bpf_map
