bpf, offload: Reorder offload callback 'prepare' in verifier
authorYinjun Zhang <yinjun.zhang@corigine.com>
Thu, 20 May 2021 08:58:34 +0000 (10:58 +0200)
committerDaniel Borkmann <daniel@iogearbox.net>
Thu, 20 May 2021 21:51:52 +0000 (23:51 +0200)
Commit 4976b718c355 ("bpf: Introduce pseudo_btf_id") switched the
order of resolve_pseudo_ldimm(), in which some pseudo instructions
are rewritten. Thus those rewritten instructions cannot be passed
to driver via 'prepare' offload callback.

Reorder the 'prepare' offload callback to fix it.

Fixes: 4976b718c355 ("bpf: Introduce pseudo_btf_id")
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20210520085834.15023-1-simon.horman@netronome.com
kernel/bpf/verifier.c

index c58598e..09849e4 100644 (file)
@@ -13368,12 +13368,6 @@ int bpf_check(struct bpf_prog **prog, union bpf_attr *attr,
        if (is_priv)
                env->test_state_freq = attr->prog_flags & BPF_F_TEST_STATE_FREQ;
 
-       if (bpf_prog_is_dev_bound(env->prog->aux)) {
-               ret = bpf_prog_offload_verifier_prep(env->prog);
-               if (ret)
-                       goto skip_full_check;
-       }
-
        env->explored_states = kvcalloc(state_htab_size(env),
                                       sizeof(struct bpf_verifier_state_list *),
                                       GFP_USER);
@@ -13401,6 +13395,12 @@ int bpf_check(struct bpf_prog **prog, union bpf_attr *attr,
        if (ret < 0)
                goto skip_full_check;
 
+       if (bpf_prog_is_dev_bound(env->prog->aux)) {
+               ret = bpf_prog_offload_verifier_prep(env->prog);
+               if (ret)
+                       goto skip_full_check;
+       }
+
        ret = check_cfg(env);
        if (ret < 0)
                goto skip_full_check;