bpf: fix sock_map_alloc() error path
authorEric Dumazet <edumazet@google.com>
Tue, 13 Feb 2018 23:33:52 +0000 (15:33 -0800)
committerAlexei Starovoitov <ast@kernel.org>
Wed, 14 Feb 2018 03:19:15 +0000 (19:19 -0800)
In case user program provides silly parameters, we want
a map_alloc() handler to return an error, not a NULL pointer,
otherwise we crash later in find_and_alloc_map()

Fixes: 1aa12bdf1bfb ("bpf: sockmap, add sock close() hook to remove socks")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
kernel/bpf/sockmap.c

index 48c33417d13c0ad40154f25aeade0c9b4cafd96a..a927e89dad6e9591066c3a87afc497a196ebd887 100644 (file)
@@ -521,8 +521,8 @@ static struct smap_psock *smap_init_psock(struct sock *sock,
 static struct bpf_map *sock_map_alloc(union bpf_attr *attr)
 {
        struct bpf_stab *stab;
-       int err = -EINVAL;
        u64 cost;
+       int err;
 
        if (!capable(CAP_NET_ADMIN))
                return ERR_PTR(-EPERM);
@@ -547,6 +547,7 @@ static struct bpf_map *sock_map_alloc(union bpf_attr *attr)
 
        /* make sure page count doesn't overflow */
        cost = (u64) stab->map.max_entries * sizeof(struct sock *);
+       err = -EINVAL;
        if (cost >= U32_MAX - PAGE_SIZE)
                goto free_stab;