Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <cover.1660761470.git.dxu@dxuuu.xyz> <edbca42217a73161903a50ba07ec63c5fa5fde00.1660761470.git.dxu@dxuuu.xyz>
In-Reply-To: <edbca42217a73161903a50ba07ec63c5fa5fde00.1660761470.git.dxu@dxuuu.xyz>
From:   Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date:   Wed, 17 Aug 2022 14:30:01 -0700
Message-ID: <CAADnVQ+G0Hju-OeN6e=JLPQzODxGXCsP7OuVbex1y-EYr6Z5Yw@mail.gmail.com>
Subject: Re: [PATCH bpf-next v2 3/4] bpf: Add support for writing to nf_conn:mark
To:     Daniel Xu <dxu@dxuuu.xyz>, Martin KaFai Lau <martin.lau@linux.dev>,
        Martin KaFai Lau <kafai@fb.com>
Cc:     bpf <bpf@vger.kernel.org>, Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Andrii Nakryiko <andrii@kernel.org>,
        Kumar Kartikeya Dwivedi <memxor@gmail.com>,
        Pablo Neira Ayuso <pablo@netfilter.org>,
        Florian Westphal <fw@strlen.de>,
        =?UTF-8?B?VG9rZSBIw7hpbGFuZC1Kw7hyZ2Vuc2Vu?= <toke@kernel.org>,
        netfilter-devel <netfilter-devel@vger.kernel.org>,
        Network Development <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Aug 17, 2022 at 11:43 AM Daniel Xu <dxu@dxuuu.xyz> wrote:
>
> +/* Check writes into `struct nf_conn` */
> +int nf_conntrack_btf_struct_access(struct bpf_verifier_log *log,
> +                                  const struct btf *btf,
> +                                  const struct btf_type *t, int off,
> +                                  int size, enum bpf_access_type atype,
> +                                  u32 *next_btf_id,
> +                                  enum bpf_type_flag *flag)
> +{
> +       const struct btf_type *nct = READ_ONCE(nf_conn_type);
> +       s32 type_id;
> +       size_t end;
> +
> +       if (!nct) {
> +               type_id = btf_find_by_name_kind(btf, "nf_conn", BTF_KIND_STRUCT);
> +               if (type_id < 0)
> +                       return -EINVAL;
> +
> +               nct = btf_type_by_id(btf, type_id);
> +               WRITE_ONCE(nf_conn_type, nct);
> +       }
> +
> +       if (t != nct) {
> +               bpf_log(log, "only read is supported\n");
> +               return -EACCES;
> +       }
> +
> +       switch (off) {
> +#if defined(CONFIG_NF_CONNTRACK_MARK)
> +       case offsetof(struct nf_conn, mark):
> +               end = offsetofend(struct nf_conn, mark);
> +               break;
> +#endif
> +       default:
> +               bpf_log(log, "no write support to nf_conn at off %d\n", off);
> +               return -EACCES;
> +       }
> +
> +       if (off + size > end) {
> +               bpf_log(log,
> +                       "write access at off %d with size %d beyond the member of nf_conn ended at %zu\n",
> +                       off, size, end);
> +               return -EACCES;
> +       }
> +
> +       return NOT_INIT;

Took me a long time to realize that this is a copy-paste
from net/ipv4/bpf_tcp_ca.c.
It's not wrong, but misleading.
When atype == BPF_READ the return value from
btf_struct_access should only be error<0, SCALAR_VALUE, PTR_TO_BTF_ID.
For atype == BPF_WRITE we should probably standardize on
error<0, or 0.

The NOT_INIT happens to be zero, but explicit 0
is cleaner to avoid confusion that this is somehow enum bpf_reg_type.

Martin,
since you've added this code in bpf_tcp_ca, wdyt?