bpf: Check sk_fullsock() before returning from bpf_sk_lookup()
authorMartin KaFai Lau <kafai@fb.com>
Fri, 17 May 2019 21:21:17 +0000 (14:21 -0700)
committerDaniel Borkmann <daniel@iogearbox.net>
Tue, 21 May 2019 14:33:30 +0000 (16:33 +0200)
commitf7355a6c049711ecfbeed573e43d48bee7acb83a
tree975e94b97c1f8b78fd7972b228b0e05027089cd4
parent9b28ae243ef3b13d8a88b5451d025475c75ebdef
bpf: Check sk_fullsock() before returning from bpf_sk_lookup()

The BPF_FUNC_sk_lookup_xxx helpers return RET_PTR_TO_SOCKET_OR_NULL.
Meaning a fullsock ptr and its fullsock's fields in bpf_sock can be
accessed, e.g. type, protocol, mark and priority.
Some new helper, like bpf_sk_storage_get(), also expects
ARG_PTR_TO_SOCKET is a fullsock.

bpf_sk_lookup() currently calls sk_to_full_sk() before returning.
However, the ptr returned from sk_to_full_sk() is not guaranteed
to be a fullsock.  For example, it cannot get a fullsock if sk
is in TCP_TIME_WAIT.

This patch checks for sk_fullsock() before returning. If it is not
a fullsock, sock_gen_put() is called if needed and then returns NULL.

Fixes: 6acc9b432e67 ("bpf: Add helper to retrieve socket in BPF")
Cc: Joe Stringer <joe@isovalent.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Joe Stringer <joe@isovalent.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
net/core/filter.c