Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Matthew Cover <matthew.cover@stackpath.com>
To:     Daniel Borkmann <daniel@iogearbox.net>,
        Alexei Starovoitov <ast@kernel.org>,
        Martin KaFai Lau <kafai@fb.com>,
        Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
        "David S. Miller" <davem@davemloft.net>,
        "netdev@vger.kernel.org" <netdev@vger.kernel.org>,
        "bpf@vger.kernel.org" <bpf@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
CC:     Matthew Cover <werekraken@gmail.com>
Subject: Re: tc_classid access in skb bpf context
Thread-Topic: tc_classid access in skb bpf context
Thread-Index: AQHVEC09Nj6ajF14HkezzoiQKGhG0aZ20sYAgACPafc=
Date:   Wed, 22 May 2019 17:26:01 +0000
Message-ID: <BYAPR10MB268073D6667D2AB0E1699D5DE3000@BYAPR10MB2680.namprd10.prod.outlook.com>
References: <BYAPR10MB2680B63C684345098E6E7669E3070@BYAPR10MB2680.namprd10.prod.outlook.com>,<73d5b951-2598-0d7f-5b6e-8925cc61989a@iogearbox.net>
In-Reply-To: <73d5b951-2598-0d7f-5b6e-8925cc61989a@iogearbox.net>
Accept-Language: en-US
Content-Language: en-US
received-spf: None (protection.outlook.com: stackpath.com does not designate
 permitted sender hosts)
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-Network-Message-Id: 44b5ec34-9049-47ef-147c-08d6deda8fda
X-MS-Exchange-CrossTenant-originalarrivaltime: 22 May 2019 17:26:01.4776
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: fd04f7e7-8712-48a5-bd2d-688fe1861f4b
X-MS-Exchange-CrossTenant-mailboxtype: HOSTED
X-MS-Exchange-Transport-CrossTenantHeadersStamped: BYAPR10MB3733
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org


On 05/22/2019 01:52 AM, Matthew Cover wrote:
> > __sk_buff has a member tc_classid which I'm interested in accessing fro=
m the skb bpf context.
> >=20
> > A bpf program which accesses skb->tc_classid compiles, but fails verifi=
cation; the specific failure is "invalid bpf_context access".
> >=20
> > if (skb->tc_classid !=3D 0)
> >=A0 return 1;
> > return 0;
> >=20
> > Some of the tests in tools/testing/selftests/bpf/verifier/ (those on tc=
_classid) further confirm that this is, in all likelihood, intentional beha=
vior.
> >=20
> > The very similar bpf program which instead accesses skb->mark works as =
desired.
> >=20
> > if (skb->mark !=3D 0)
> >=A0 return 1;
> > return 0;
>=20
> You should be able to access skb->tc_classid, perhaps you're using the wr=
ong program
> type? BPF_PROG_TYPE_SCHED_CLS is supposed to work (if not we'd have a reg=
ression).
>=20

I am in fact using BPF_PROG_TYPE_SOCKET_FILTER and using the program as PAC=
KET_FANOUT_DATA with PACKET_FANOUT_EBPF.

I have been working on a series of utils which leverage PACKET_FANOUT to pr=
ovide various per-socket-fd (per-cpu, per-queue, per-rx-flow-hash-indirecti=
on-table-idx) statistics and pcap files. While playing with PACKET_FANOUT_E=
BPF, I realized that I could use the bpf program to categorize packets in w=
ays packet-filter(7) does not provide.

As a concrete example, I plan to build a util `rxtxmark` which could be pas=
sed something like `--mark-list 42,88`. This would be translated to a bpf p=
rogram where the return code is the ordinality of the mark in the list.

if (skb->mark =3D=3D 42)
 return 1;
if (skb->mark =3D=3D 88)
 return 2;
return 0;

Packets enqueued to fd0 are simply ignored. Packets enqueued to the other f=
ds are processed into pcaps and statistics.

While I may build a util for tc_classid which does per-user-requested-class=
id pcaps and statistics like `rxtxmark` does for marks, I'm also interested=
 in using tc_classid as a simple way to capture tx packets from a long runn=
ing program on the fly.

The program under inspection would simply be added to a net_cls cgroup whic=
h has a unique classid defined. A bpf program would be attached to map pack=
ets with that classid to fd1. While I can do this already by using iptables=
 to translate the tc_classid to a mark, that complicates the implementation=
 greatly since the firewall has to be touched (which is probably overreachi=
ng for a packet capture util and would most likely be left to the user to c=
onfigure).

> > I built a kernel (v5.1) with 4 instances of the following line removed =
from net/core/filter.c to test the behavior when the instructions pass veri=
fication.
> >=20
> >=A0=A0=A0=A0 switch (off) {
> > -=A0=A0=A0 case bpf_ctx_range(struct __sk_buff, tc_classid):
> > ...
> >=A0=A0=A0=A0=A0=A0=A0=A0 return false;
> >=20
> > It appears skb->tc_classid is always zero within my bpf program, even w=
hen I verify by other means (e.g. netfilter) that the value is set non-zero=
.
> >=20
> > I gather that sk_buff proper sometimes (i.e. at some layers) has qdisc_=
skb_cb stored in skb->cb, but not always.
> >=20
> > I suspect that the tc_classid is available at l3 (and therefore to util=
s like netfilter, ip route, tc), but not at l2 (and not to AF_PACKET).
>=20
> From tc/BPF context you can use it; it's been long time, but I think back=
 then
> we mapped it into cb[] so it can be used within the BPF context to pass s=
kb data
> around e.g. between tail calls, and cls_bpf_classify() when in direct-act=
ion mode
> which likely everyone is/should-be using then maps that skb->tc_classid u=
16 cb[]
> value to res->classid on program return which then in either sch_handle_i=
ngress()
> or sch_handle_egress() is transferred into the skb->tc_index.
>=20

It sounds like just before the start of a BPF_PROG_TYPE_SCHED_CLS bpf progr=
am tc_classid id placed in skb->cb. The missing plumbing to support my use =
case is probably the same thing, but for BPF_PROG_TYPE_SOCKET_FILTER.

I'll see about familiarizing myself with both as time permits and perhaps I=
 can get tc_classid working for a BPF_PROG_TYPE_SOCKET_FILTER program; it c=
ertainly sounds like it's doable.

> > Is it impractical to make skb->tc_classid available in this bpf context=
 or is there just some plumbing which hasn't been connected yet?
> >=20
> > Is my suspicion that skb->cb no longer contains qdisc_skb_cb due to cro=
ssing a layer boundary well founded?
> >=20
> > I'm willing to look into hooking things together as time permits if it'=
s a feasible task.
> >=20
> > It's trivial to have iptables match on tc_classid and set a mark which =
is available to bpf at l2, but I'd like to better understand this.
> >=20
> > Thanks,
> > Matt C.
> >=20

    =