Re: Automated report: NetBSD-current/i386 test failure

04.09.2018, 16:38, "NetBSD Test Fixture" <bracket%netbsd.org@localhost>:

This is an automatically generated notice of new failures of the
NetBSD test suite.

The newly failing test cases are:

    lib/libbpfjit/t_bpfjit:libbpfjit_bad_ret_k
    lib/libbpfjit/t_bpfjit:libbpfjit_cop_no_ctx
    lib/libbpfjit/t_bpfjit:libbpfjit_copx_no_ctx
    lib/libbpfjit/t_cop:libbpfjit_cop_no_ctx
    lib/libbpfjit/t_cop:libbpfjit_copx_no_ctx

The above tests failed in each of the last 3 test runs, and passed in
at least 27 consecutive runs before that.

These failures aren’t in bpfjit but in bpf_validate function. Most likely, the new libpcap understands BPF_COP and it accepts it as a valid instruction while my tests assume that BPF_COP is only valid if it has been registered before the validate call.

Between the last successful test and the failed test, a total of 700
revisions were committed, by the following developers:

    christos
    kre
    riastradh

Main suspect is libpcap import done by christos.

Log files can be found at:

http://releng.NetBSD.org/b5reports/i386/commits-2018.09.html#2018.09.03.21.29.30

Alex