bpf: Enforce returning 0 for fentry/fexit progs
authorYonghong Song <yhs@fb.com>
Thu, 14 May 2020 05:32:05 +0000 (22:32 -0700)
committerAlexei Starovoitov <ast@kernel.org>
Thu, 14 May 2020 19:50:10 +0000 (12:50 -0700)
Currently, tracing/fentry and tracing/fexit prog
return values are not enforced. In trampoline codes,
the fentry/fexit prog return values are ignored.
Let us enforce it to be 0 to avoid confusion and
allows potential future extension.

This patch also explicitly added return value
checking for tracing/raw_tp, tracing/fmod_ret,
and freplace programs such that these program
return values can be anything. The purpose are
two folds:
 1. to make it explicit about return value expectations
    for these programs in verifier.
 2. for tracing prog_type, if a future attach type
    is added, the default is -ENOTSUPP which will
    enforce to specify return value ranges explicitly.

Fixes: fec56f5890d9 ("bpf: Introduce BPF trampoline")
Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/20200514053206.1298415-1-yhs@fb.com
kernel/bpf/verifier.c

index fa1d8245b9257fe6f0f40dfca35702ebe7662000..a44ba66726882539895f7550487cf30aefc3ff82 100644 (file)
@@ -7059,6 +7059,23 @@ static int check_return_code(struct bpf_verifier_env *env)
                        return 0;
                range = tnum_const(0);
                break;
+       case BPF_PROG_TYPE_TRACING:
+               switch (env->prog->expected_attach_type) {
+               case BPF_TRACE_FENTRY:
+               case BPF_TRACE_FEXIT:
+                       range = tnum_const(0);
+                       break;
+               case BPF_TRACE_RAW_TP:
+               case BPF_MODIFY_RETURN:
+                       return 0;
+               default:
+                       return -ENOTSUPP;
+               }
+               break;
+       case BPF_PROG_TYPE_EXT:
+               /* freplace program can return anything as its return value
+                * depends on the to-be-replaced kernel func or bpf program.
+                */
        default:
                return 0;
        }