Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Song Liu <songliubraving@fb.com>
To:     Andrii Nakryiko <andrii.nakryiko@gmail.com>
CC:     bpf <bpf@vger.kernel.org>, Networking <netdev@vger.kernel.org>,
        open list <linux-kernel@vger.kernel.org>,
        Alexei Starovoitov <ast@kernel.org>,
        "Daniel Borkmann" <daniel@iogearbox.net>,
        Kernel Team <Kernel-team@fb.com>,
        "Peter Ziljstra" <peterz@infradead.org>
Subject: Re: [PATCH v4 bpf-next 2/6] bpf: prevent deadlock from recursive
 bpf_task_storage_[get|delete]
Thread-Topic: [PATCH v4 bpf-next 2/6] bpf: prevent deadlock from recursive
 bpf_task_storage_[get|delete]
Thread-Index: AQHXCYIynt1KuoobvEWNWRDMRSL2HaplRT0AgAAPSwA=
Date:   Tue, 23 Feb 2021 07:16:18 +0000
Message-ID: <6A4F1927-AF73-4AC8-AE44-5878ACEDF944@fb.com>
References: <20210223012014.2087583-1-songliubraving@fb.com>
 <20210223012014.2087583-3-songliubraving@fb.com>
 <CAEf4BzaZ0ATbJsLoQu_SRUYgzkak9zv61N+T=gijOQ+X=57ErA@mail.gmail.com>
In-Reply-To: <CAEf4BzaZ0ATbJsLoQu_SRUYgzkak9zv61N+T=gijOQ+X=57ErA@mail.gmail.com>
Accept-Language: en-US
Content-Language: en-US
Content-Type: text/plain; charset="us-ascii"
Content-ID: <56FFB24331712B43B00CE1763F75904D@namprd15.prod.outlook.com>
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-Network-Message-Id: 948c1cd3-10d6-400d-537d-08d8d7caea40
X-MS-Exchange-CrossTenant-originalarrivaltime: 23 Feb 2021 07:16:18.4885
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: 8ae927fe-1255-47a7-a2af-5f3a069daaa2
X-MS-Exchange-CrossTenant-mailboxtype: HOSTED
X-MS-Exchange-CrossTenant-userprincipalname: vgLu5HY+WNU6SB+RVQyuH843rr8lXAj4mr1FEVglW5RVBYscNvt3aeyIO9SIzHiEspQcavQXKv8IuRgxXOFH6g==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: SJ0PR15MB4711
X-OriginatorOrg: fb.com
X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10434:6.0.369,18.0.761
 definitions=2021-02-23_02:2021-02-22,2021-02-23 signatures=0
X-Proofpoint-Spam-Details: rule=fb_default_notspam policy=fb_default score=0 mlxscore=0 adultscore=0
 malwarescore=0 bulkscore=0 spamscore=0 suspectscore=0 priorityscore=1501
 phishscore=0 impostorscore=0 clxscore=1015 mlxlogscore=999
 lowpriorityscore=0 classifier=spam adjust=0 reason=mlx scancount=1
 engine=8.12.0-2009150000 definitions=main-2102230059
X-FB-Internal: deliver
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org


> On Feb 22, 2021, at 10:21 PM, Andrii Nakryiko <andrii.nakryiko@gmail.com>=
 wrote:
>=20
> On Mon, Feb 22, 2021 at 5:23 PM Song Liu <songliubraving@fb.com> wrote:
>>=20
>> BPF helpers bpf_task_storage_[get|delete] could hold two locks:
>> bpf_local_storage_map_bucket->lock and bpf_local_storage->lock. Calling
>> these helpers from fentry/fexit programs on functions in bpf_*_storage.c
>> may cause deadlock on either locks.
>>=20
>> Prevent such deadlock with a per cpu counter, bpf_task_storage_busy, whi=
ch
>> is similar to bpf_prog_active. We need this counter to be global, becaus=
e
>> the two locks here belong to two different objects: bpf_local_storage_ma=
p
>> and bpf_local_storage. If we pick one of them as the owner of the counte=
r,
>> it is still possible to trigger deadlock on the other lock. For example,
>> if bpf_local_storage_map owns the counters, it cannot prevent deadlock
>> on bpf_local_storage->lock when two maps are used.
>>=20
>> Signed-off-by: Song Liu <songliubraving@fb.com>
>> ---
>> kernel/bpf/bpf_task_storage.c | 57 ++++++++++++++++++++++++++++++-----
>> 1 file changed, 50 insertions(+), 7 deletions(-)
>>=20
>=20
> [...]
>=20
>> @@ -109,7 +136,9 @@ static void *bpf_pid_task_storage_lookup_elem(struct=
 bpf_map *map, void *key)
>>                goto out;
>>        }
>>=20
>> +       bpf_task_storage_lock();
>>        sdata =3D task_storage_lookup(task, map, true);
>> +       bpf_task_storage_unlock();
>>        put_pid(pid);
>>        return sdata ? sdata->data : NULL;
>> out:
>> @@ -141,8 +170,10 @@ static int bpf_pid_task_storage_update_elem(struct =
bpf_map *map, void *key,
>>                goto out;
>>        }
>>=20
>> +       bpf_task_storage_lock();
>>        sdata =3D bpf_local_storage_update(
>>                task, (struct bpf_local_storage_map *)map, value, map_fla=
gs);
>=20
> this should probably be container_of() instead of casting

bpf_task_storage.c uses casting in multiple places. How about we fix it in =
a=20
separate patch?

Thanks,
Song

>=20
>> +       bpf_task_storage_unlock();
>>=20
>>        err =3D PTR_ERR_OR_ZERO(sdata);
>> out:
>> @@ -185,7 +216,9 @@ static int bpf_pid_task_storage_delete_elem(struct b=
pf_map *map, void *key)
>>                goto out;
>>        }
>>=20
>> +       bpf_task_storage_lock();
>>        err =3D task_storage_delete(task, map);
>> +       bpf_task_storage_unlock();
>> out:
>>        put_pid(pid);
>>        return err;
>=20
> [...]