On 05-Mär 09:38, Alexei Starovoitov wrote: > On Thu, Mar 5, 2020 at 9:32 AM KP Singh <kpsingh@xxxxxxxxxxxx> wrote: > > > > This fails as we added bpf_test_run_tracing in net/bpf/test_run.c > > which gets built only CONFIG_NET is enabled. Which, this particular > > config, disables. > > > > Alexei, if it's okay with you. I can send a patch that separates the > > tracing test code into kernel/bpf/test_run_trace.c which depends > > only on CONFIG_BPF_SYSCALL. > > In such situation we typically add __weak dummy call. I would prefer this. Less chances for breaking something. Sent: https://lore.kernel.org/bpf/20200305220127.29109-1-kpsingh@xxxxxxxxxxxx/T/#u > May be split will work too. We can do that separately (if needed). - KP > or move tracing_prog_ops to kernel/bpf/core.c ?
