diff options
| author | Daniel Borkmann <daniel@iogearbox.net> | 2018-06-15 02:30:47 +0200 |
|---|---|---|
| committer | Alexei Starovoitov <ast@kernel.org> | 2018-06-15 11:14:25 -0700 |
| commit | 7d1982b4e335c1b184406b7566f6041bfe313c35 (patch) | |
| tree | 885f83b1f8e96c502c2939838d22e63756c7b011 /kernel/bpf/core.c | |
| parent | 26bf8a89d887c0686acef0f44eaadd49abfcab03 (diff) | |
bpf: fix panic in prog load calls cleanup
While testing I found that when hitting error path in bpf_prog_load()
where we jump to free_used_maps and prog contained BPF to BPF calls
that were JITed earlier, then we never clean up the bpf_prog_kallsyms_add()
done under jit_subprogs(). Add proper API to make BPF kallsyms deletion
more clear and fix that.
Fixes: 1c2a088a6626 ("bpf: x64: add JIT support for multi-function programs")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Diffstat (limited to 'kernel/bpf/core.c')
| -rw-r--r-- | kernel/bpf/core.c | 14 |
1 files changed, 14 insertions, 0 deletions
diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index 9f1493705f40..1061968adcc1 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -350,6 +350,20 @@ struct bpf_prog *bpf_patch_insn_single(struct bpf_prog *prog, u32 off, return prog_adj; } +void bpf_prog_kallsyms_del_subprogs(struct bpf_prog *fp) +{ + int i; + + for (i = 0; i < fp->aux->func_cnt; i++) + bpf_prog_kallsyms_del(fp->aux->func[i]); +} + +void bpf_prog_kallsyms_del_all(struct bpf_prog *fp) +{ + bpf_prog_kallsyms_del_subprogs(fp); + bpf_prog_kallsyms_del(fp); +} + #ifdef CONFIG_BPF_JIT /* All BPF JIT sysctl knobs here. */ int bpf_jit_enable __read_mostly = IS_BUILTIN(CONFIG_BPF_JIT_ALWAYS_ON); |