[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200922201600.5cz6vslfiuw7hdkw@ast-mbp.dhcp.thefacebook.com>
Date: Tue, 22 Sep 2020 13:16:00 -0700
From: Alexei Starovoitov <alexei.starovoitov@...il.com>
To: Martin KaFai Lau <kafai@...com>
Cc: bpf@...r.kernel.org, Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>, kernel-team@...com,
Lorenz Bauer <lmb@...udflare.com>, netdev@...r.kernel.org
Subject: Re: [PATCH v3 bpf-next 11/11] bpf: selftest: Use
bpf_skc_to_tcp_sock() in the sock_fields test
On Tue, Sep 22, 2020 at 12:05:18AM -0700, Martin KaFai Lau wrote:
>
> @@ -165,9 +168,13 @@ int egress_read_sock_fields(struct __sk_buff *skb)
> tpcpy(tp_ret, tp);
>
> if (sk_ret == &srv_sk) {
> + ktp = bpf_skc_to_tcp_sock(sk);
> + if (!ktp)
> + RET_LOG();
> + lsndtime = ktp->lsndtime;
If I understood the patches correctly they extend the use cases significantly,
but this test only covers cgroup/egress type with one particular helper.
Could you please expand the test cases to cover more of things from earlier
patches. Patch 7 covers some, but imo not enough.
Similarly the case of sk_release(sk) in one place and sk_release(ktp)
in another branch (one you mention in commit log) needs to have
C based selftest as well.
Powered by blists - more mailing lists