[PATCH mptcp-next v5 0/5] Squash to "Add mptcp_subflow bpf_iter support"
[PATCH mptcp-next v5 0/5] Squash to "Add mptcp_subflow bpf_iter support"
1
From: Geliang Tang <tanggeliang@kylinos.cn>
1
From: Geliang Tang <tanggeliang@kylinos.cn>
2
2
3
v5:
3
v5:
4
- update subjects and commit logs as Matt suggested (thanks).
4
- check bpf_iter_task in mptcp_subflow_new() as Mat suggested.
5
- avoid repeating checks in patch 1.
6
- do stricter checks on subflows in patch 3.
7
5
8
v4:
6
v4:
9
- CI reports the following BUILD_BUG_ON fails on i386:
7
- drop sock_owned_by_user_nocheck and spin_is_locked. According to
10
8
comments from Mat and Martin, in this set mptcp_subflow
11
BUILD_BUG_ON(sizeof(struct bpf_iter_mptcp_subflow_kern) !=
9
bpf_iter only used from a cg sockopt bpf prog, no need to add these
12
     sizeof(struct bpf_iter_mptcp_subflow))
10
check at this moment.
13
14
Just like in bpf_iter_task_new(), change this "!=" to ">".
15
11
16
v3:
12
v3:
17
- check sock_owned_by_user_nocheck(sk)/spin_is_locked(&sk->sk_lock.slock),
13
- patch 3, continue to use sock_owned_by_user_nocheck() and spin_is_locked()
18
instead of lockdep_sock_is_held(sk).
14
checks instead of using msk_owned_by_me().
19
- add "sizeof" and "alignof" checks.
15
- patch 5, drop declaration of bpf_mptcp_subflow_tcp_sock. It's no longer
20
- drop bpf_mptcp_sk() and bpf_mptcp_subflow_tcp_sock() definitions. Use
16
used.
21
bpf_skc_to_mptcp_sock() and mptcp_subflow_tcp_sock() in mptcp_subflow
17
- patch 5, update the comment for mptcp_subflow_tcp_sock(), which is a BPF
22
bpf_iter selftests instead.
18
helper, not a kfunc.
19
20
The commit log of "bpf: Register mptcp common kfunc set" doesn't match the
21
code, please update it as:
22
23
'''
24
bpf: Register mptcp common kfunc set
25
26
MPTCP helper mptcp_subflow_ctx() is used to convert struct sock to
27
struct mptcp_subflow_context. It will be used in MPTCP BPF programs.
28
29
This patch defines corresponding wrapper of this helper, and put it
30
into the newly defined mptcp common kfunc set and register this set
31
with the flag BPF_PROG_TYPE_CGROUP_SOCKOPT to let it accessible to
32
the 'cgroup/getsockopt' type of BPF programs.
33
'''
23
34
24
v2:
35
v2:
25
- add CONFIG_LOCKDEP check in patch 2 to fix the build error reported
36
- Drop bpf_skc_to_mptcp_sock
26
by CI.
37
- Check the owner before assigning the msk as Mat suggested.
38
- Use bpf_core_cast() in mptcp_subflow bpf_iter subtest instead of
39
using bpf_skc_to_mptcp_sock().
27
40
28
Address Martin's comments in v1.
41
Address Martin's suggestions for "Add mptcp_subflow bpf_iter support" v2.
29
42
30
Geliang Tang (5):
43
Geliang Tang (5):
31
bpf: Extend bpf_skc_to_mptcp_sock to MPTCP sock
44
mptcp: add bpf_iter_task for mptcp_sock
32
bpf: Allow use of skc_to_mptcp_sock in cg_sockopt
45
Squash to "bpf: Extend bpf_skc_to_mptcp_sock to MPTCP sock"
33
Squash to "bpf: Register mptcp common kfunc set"
34
Squash to "bpf: Add mptcp_subflow bpf_iter"
46
Squash to "bpf: Add mptcp_subflow bpf_iter"
47
Revert "bpf: Acquire and release mptcp socket"
35
Squash to "selftests/bpf: Add mptcp_subflow bpf_iter subtest"
48
Squash to "selftests/bpf: Add mptcp_subflow bpf_iter subtest"
36
49
37
include/net/mptcp.h | 4 +-
50
net/mptcp/bpf.c | 56 +++++++++----------
38
kernel/bpf/cgroup.c | 2 +
51
net/mptcp/protocol.c | 1 +
39
net/core/filter.c | 2 +-
52
net/mptcp/protocol.h | 16 ++++++
40
net/mptcp/bpf.c | 41 +++++++++++--------
53
.../testing/selftests/bpf/bpf_experimental.h | 2 +-
41
tools/testing/selftests/bpf/progs/mptcp_bpf.h | 1 -
54
tools/testing/selftests/bpf/progs/mptcp_bpf.h | 5 --
42
.../selftests/bpf/progs/mptcp_bpf_iters.c | 11 +++--
55
.../selftests/bpf/progs/mptcp_bpf_iters.c | 8 +--
43
6 files changed, 33 insertions(+), 28 deletions(-)
56
6 files changed, 47 insertions(+), 41 deletions(-)
44
57
45
--
58
--
46
2.45.2
59
2.43.0
diff view generated by jsdifflib
[PATCH mptcp-next v5 2/5] bpf: Allow use of skc_to_mptcp_sock in cg_sockopt
[PATCH mptcp-next v5 1/5] mptcp: add bpf_iter_task for mptcp_sock
1
From: Geliang Tang <tanggeliang@kylinos.cn>
1
From: Geliang Tang <tanggeliang@kylinos.cn>
2
2
3
Currently, bpf_skc_to_mptcp_sock() helper is not allowed to be used
3
To make sure the mptcp_subflow bpf_iter is running in the
4
in cg_sockopt. This patch adds this permission.
4
MPTCP context. This patch adds a simplified version of tracking
5
for it:
5
6
6
Thanks to the previous patch allowing skc_to_mptcp_sock() to be used
7
1. Add a 'struct task_struct *bpf_iter_task' field to struct
7
with MPTCP sockets, this permission allows this helper to be use it in
8
mptcp_sock.
8
CGroup BPF hooks, e.g. [gs]etsocktopt.
9
9
10
2. Do a WRITE_ONCE(msk->bpf_iter_task, current) before calling
11
a MPTCP BPF hook, and WRITE_ONCE(msk->bpf_iter_task, NULL) after
12
the hook returns.
13
14
3. In bpf_iter_mptcp_subflow_new(), check
15
16
    "READ_ONCE(msk->bpf_scheduler_task) == current"
17
18
to confirm the correct task, return -EINVAL if it doesn't match.
19
20
Also creates helpers for setting, clearing and checking that value.
21
22
Suggested-by: Mat Martineau <martineau@kernel.org>
10
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
23
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
11
---
24
---
12
kernel/bpf/cgroup.c | 2 ++
25
net/mptcp/protocol.c | 1 +
13
1 file changed, 2 insertions(+)
26
net/mptcp/protocol.h | 16 ++++++++++++++++
27
2 files changed, 17 insertions(+)
14
28
15
diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
29
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
16
index XXXXXXX..XXXXXXX 100644
30
index XXXXXXX..XXXXXXX 100644
17
--- a/kernel/bpf/cgroup.c
31
--- a/net/mptcp/protocol.c
18
+++ b/kernel/bpf/cgroup.c
32
+++ b/net/mptcp/protocol.c
19
@@ -XXX,XX +XXX,XX @@ cg_sockopt_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
33
@@ -XXX,XX +XXX,XX @@ static void __mptcp_init_sock(struct sock *sk)
20
#ifdef CONFIG_INET
34
    msk->scaling_ratio = TCP_DEFAULT_SCALING_RATIO;
21
    case BPF_FUNC_tcp_sock:
35
22
        return &bpf_tcp_sock_proto;
36
    WRITE_ONCE(msk->first, NULL);
23
+    case BPF_FUNC_skc_to_mptcp_sock:
37
+    WRITE_ONCE(msk->bpf_iter_task, NULL);
24
+        return &bpf_skc_to_mptcp_sock_proto;
38
    inet_csk(sk)->icsk_sync_mss = mptcp_sync_mss;
39
    WRITE_ONCE(msk->csum_enabled, mptcp_is_checksum_enabled(sock_net(sk)));
40
    WRITE_ONCE(msk->allow_infinite_fallback, true);
41
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
42
index XXXXXXX..XXXXXXX 100644
43
--- a/net/mptcp/protocol.h
44
+++ b/net/mptcp/protocol.h
45
@@ -XXX,XX +XXX,XX @@ struct mptcp_sock {
46
    struct list_head conn_list;
47
    struct list_head rtx_queue;
48
    struct mptcp_data_frag *first_pending;
49
+    struct task_struct *bpf_iter_task;
50
    struct list_head join_list;
51
    struct sock    *first; /* The mptcp ops can safely dereference, using suitable
52
                 * ONCE annotation, the subflow outside the socket
53
@@ -XXX,XX +XXX,XX @@ mptcp_token_join_cookie_init_state(struct mptcp_subflow_request_sock *subflow_re
54
static inline void mptcp_join_cookie_init(void) {}
25
#endif
55
#endif
26
    case BPF_FUNC_perf_event_output:
56
27
        return &bpf_event_output_data_proto;
57
+static inline void mptcp_set_bpf_iter_task(struct mptcp_sock *msk)
58
+{
59
+    WRITE_ONCE(msk->bpf_iter_task, current);
60
+}
61
+
62
+static inline void mptcp_clear_bpf_iter_task(struct mptcp_sock *msk)
63
+{
64
+    WRITE_ONCE(msk->bpf_iter_task, NULL);
65
+}
66
+
67
+static inline struct task_struct *mptcp_get_bpf_iter_task(struct mptcp_sock *msk)
68
+{
69
+    return READ_ONCE(msk->bpf_iter_task);
70
+}
71
+
72
#endif /* __MPTCP_PROTOCOL_H */
28
--
73
--
29
2.45.2
74
2.43.0
diff view generated by jsdifflib
[PATCH mptcp-next v5 1/5] bpf: Extend bpf_skc_to_mptcp_sock to MPTCP sock
[PATCH mptcp-next v5 2/5] Squash to "bpf: Extend bpf_skc_to_mptcp_sock to MPTCP sock"
1
From: Geliang Tang <tanggeliang@kylinos.cn>
1
From: Geliang Tang <tanggeliang@kylinos.cn>
2
2
3
Currently, bpf_skc_to_mptcp_sock() can only be used with sockets that
3
Set msk->bpf_iter_task in bpf_mptcp_sock_from_sock() to allow
4
are MPTCP subflows: TCP sockets with tp->is_mptcp, created by the kernel
4
mptcp_subflow bpt_iter can be used in cgroup/getsockopt,
5
from an MPTCP socket (IPPROTO_MPTCP). Typically used with BPF sock_ops
5
otherwise, the selftest in this set fails.
6
operators.
7
8
Here, this helper is extended to support MPTCP sockets, the ones created
9
by the userspace (IPPROTO_MPTCP). This is useful for BPF hooks involving
10
these sockets, e.g. [gs]etsocktopt.
11
12
bpf_skc_to_mptcp_sock() uses bpf_mptcp_sock_from_subflow(). The former
13
suggests any MPTCP type/subtype can be used, but the latter only accepts
14
subflow ones. So bpf_mptcp_sock_from_subflow is modified here to support
15
MPTCP socket, and renamed to avoid confusions.
16
6
17
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
7
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
18
---
8
---
19
include/net/mptcp.h | 4 ++--
9
net/mptcp/bpf.c | 16 ++++++++++++----
20
net/core/filter.c | 2 +-
10
1 file changed, 12 insertions(+), 4 deletions(-)
21
net/mptcp/bpf.c | 10 ++++++++--
22
3 files changed, 11 insertions(+), 5 deletions(-)
23
11
24
diff --git a/include/net/mptcp.h b/include/net/mptcp.h
25
index XXXXXXX..XXXXXXX 100644
26
--- a/include/net/mptcp.h
27
+++ b/include/net/mptcp.h
28
@@ -XXX,XX +XXX,XX @@ static inline void mptcpv6_handle_mapped(struct sock *sk, bool mapped) { }
29
#endif
30
31
#if defined(CONFIG_MPTCP) && defined(CONFIG_BPF_SYSCALL)
32
-struct mptcp_sock *bpf_mptcp_sock_from_subflow(struct sock *sk);
33
+struct mptcp_sock *bpf_mptcp_sock_from_sock(struct sock *sk);
34
#else
35
-static inline struct mptcp_sock *bpf_mptcp_sock_from_subflow(struct sock *sk) { return NULL; }
36
+static inline struct mptcp_sock *bpf_mptcp_sock_from_sock(struct sock *sk) { return NULL; }
37
#endif
38
39
#if !IS_ENABLED(CONFIG_MPTCP)
40
diff --git a/net/core/filter.c b/net/core/filter.c
41
index XXXXXXX..XXXXXXX 100644
42
--- a/net/core/filter.c
43
+++ b/net/core/filter.c
44
@@ -XXX,XX +XXX,XX @@ const struct bpf_func_proto bpf_skc_to_unix_sock_proto = {
45
BPF_CALL_1(bpf_skc_to_mptcp_sock, struct sock *, sk)
46
{
47
    BTF_TYPE_EMIT(struct mptcp_sock);
48
-    return (unsigned long)bpf_mptcp_sock_from_subflow(sk);
49
+    return (unsigned long)bpf_mptcp_sock_from_sock(sk);
50
}
51
52
const struct bpf_func_proto bpf_skc_to_mptcp_sock_proto = {
53
diff --git a/net/mptcp/bpf.c b/net/mptcp/bpf.c
12
diff --git a/net/mptcp/bpf.c b/net/mptcp/bpf.c
54
index XXXXXXX..XXXXXXX 100644
13
index XXXXXXX..XXXXXXX 100644
55
--- a/net/mptcp/bpf.c
14
--- a/net/mptcp/bpf.c
56
+++ b/net/mptcp/bpf.c
15
+++ b/net/mptcp/bpf.c
57
@@ -XXX,XX +XXX,XX @@ static struct bpf_struct_ops bpf_mptcp_sched_ops = {
16
@@ -XXX,XX +XXX,XX @@ static struct bpf_struct_ops bpf_mptcp_sched_ops = {
58
};
17
59
#endif /* CONFIG_BPF_JIT */
18
struct mptcp_sock *bpf_mptcp_sock_from_sock(struct sock *sk)
60
61
-struct mptcp_sock *bpf_mptcp_sock_from_subflow(struct sock *sk)
62
+struct mptcp_sock *bpf_mptcp_sock_from_sock(struct sock *sk)
63
{
19
{
64
-    if (sk && sk_fullsock(sk) && sk->sk_protocol == IPPROTO_TCP && sk_is_mptcp(sk))
20
+    struct mptcp_sock *msk;
65
+    if (unlikely(!sk || !sk_fullsock(sk)))
66
+        return NULL;
67
+
21
+
68
+    if (sk->sk_protocol == IPPROTO_MPTCP)
22
    if (unlikely(!sk || !sk_fullsock(sk)))
69
+        return mptcp_sk(sk);
23
        return NULL;
70
+
24
71
+    if (sk->sk_protocol == IPPROTO_TCP && sk_is_mptcp(sk))
25
-    if (sk->sk_protocol == IPPROTO_MPTCP)
72
        return mptcp_sk(mptcp_subflow_ctx(sk)->conn);
26
-        return mptcp_sk(sk);
27
+    if (sk->sk_protocol == IPPROTO_MPTCP) {
28
+        msk = mptcp_sk(sk);
29
+        mptcp_set_bpf_iter_task(msk);
30
+        return msk;
31
+    }
32
33
-    if (sk->sk_protocol == IPPROTO_TCP && sk_is_mptcp(sk))
34
-        return mptcp_sk(mptcp_subflow_ctx(sk)->conn);
35
+    if (sk->sk_protocol == IPPROTO_TCP && sk_is_mptcp(sk)) {
36
+        msk = mptcp_sk(mptcp_subflow_ctx(sk)->conn);
37
+        mptcp_set_bpf_iter_task(msk);
38
+        return msk;
39
+    }
73
40
74
    return NULL;
41
    return NULL;
42
}
75
--
43
--
76
2.45.2
44
2.43.0
diff view generated by jsdifflib
[PATCH mptcp-next v5 4/5] Squash to "bpf: Add mptcp_subflow bpf_iter"
[PATCH mptcp-next v5 3/5] Squash to "bpf: Add mptcp_subflow bpf_iter"
1
From: Geliang Tang <tanggeliang@kylinos.cn>
1
From: Geliang Tang <tanggeliang@kylinos.cn>
2
2
3
Add "sizeof" and "alignof" checks.
3
Drop the NULL check for 'msk' as Martin suggested, add more checks
4
for 'sk'.
4
5
5
Address Martin's comments in v1:
6
Use the "struct sock *sk" instead of "struct mptcp-sock *msk" as the
7
argument in the bpf_iter_mptcp_subflow_new as Martin suggested.
6
8
7
- bpf_iter_mptcp_subflow_new returns -EINVAL when msk socket lock isn't
9
v5:
8
held.
10
- check bpf_iter_task in mptcp_subflow_new()
11
12
v4:
13
- drop sock_owned_by_user_nocheck and spin_is_locked. According to
14
comments from Mat [2] and Martin [1], in this set mptcp_subflow
15
bpf_iter only used from a cg sockopt bpf prog, no need to add these
16
check at this moment.
17
18
[1]
19
https://lore.kernel.org/all/fdf0ddbe-e007-4a5f-bbdf-9a144e8fbe35@linux.dev/
20
[2]
21
https://patchwork.kernel.org/project/mptcp/patch/f6469225598beecbf0bda12a4c33fafa86c0ff15.1739787744.git.tanggeliang@kylinos.cn/
22
23
v3:
24
- continue to use sock_owned_by_user_nocheck and spin_is_locked
25
checks instead of using msk_owned_by_me().
26
27
v2:
28
- check the owner before assigning the msk as Mat suggested.
9
29
10
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
30
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
11
---
31
---
12
net/mptcp/bpf.c | 10 +++++++++-
32
net/mptcp/bpf.c | 21 +++++++++++++++------
13
1 file changed, 9 insertions(+), 1 deletion(-)
33
1 file changed, 15 insertions(+), 6 deletions(-)
14
34
15
diff --git a/net/mptcp/bpf.c b/net/mptcp/bpf.c
35
diff --git a/net/mptcp/bpf.c b/net/mptcp/bpf.c
16
index XXXXXXX..XXXXXXX 100644
36
index XXXXXXX..XXXXXXX 100644
17
--- a/net/mptcp/bpf.c
37
--- a/net/mptcp/bpf.c
18
+++ b/net/mptcp/bpf.c
38
+++ b/net/mptcp/bpf.c
19
@@ -XXX,XX +XXX,XX @@ bpf_iter_mptcp_subflow_new(struct bpf_iter_mptcp_subflow *it,
39
@@ -XXX,XX +XXX,XX @@ bpf_mptcp_subflow_ctx(const struct sock *sk)
20
             struct mptcp_sock *msk)
40
41
__bpf_kfunc static int
42
bpf_iter_mptcp_subflow_new(struct bpf_iter_mptcp_subflow *it,
43
-             struct mptcp_sock *msk)
44
+             struct sock *sk)
21
{
45
{
22
    struct bpf_iter_mptcp_subflow_kern *kit = (void *)it;
46
    struct bpf_iter_mptcp_subflow_kern *kit = (void *)it;
23
+    struct sock *sk = (struct sock *)msk;
47
-    struct sock *sk = (struct sock *)msk;
48
+    struct task_struct *task;
49
+    struct mptcp_sock *msk;
50
51
    BUILD_BUG_ON(sizeof(struct bpf_iter_mptcp_subflow_kern) >
52
         sizeof(struct bpf_iter_mptcp_subflow));
53
    BUILD_BUG_ON(__alignof__(struct bpf_iter_mptcp_subflow_kern) !=
54
         __alignof__(struct bpf_iter_mptcp_subflow));
55
56
-    kit->msk = msk;
57
-    if (!msk)
58
+    if (unlikely(!sk || !sk_fullsock(sk)))
59
        return -EINVAL;
60
61
-    if (!sock_owned_by_user_nocheck(sk) &&
62
-     !spin_is_locked(&sk->sk_lock.slock))
63
+    if (sk->sk_protocol != IPPROTO_MPTCP)
64
        return -EINVAL;
65
66
+    msk = mptcp_sk(sk);
67
+    task = mptcp_get_bpf_iter_task(msk);
68
+    if (!task || task != current)
69
+        return -EINVAL;
24
+
70
+
25
+    BUILD_BUG_ON(sizeof(struct bpf_iter_mptcp_subflow_kern) >
71
+    mptcp_clear_bpf_iter_task(msk);
26
+         sizeof(struct bpf_iter_mptcp_subflow));
72
+
27
+    BUILD_BUG_ON(__alignof__(struct bpf_iter_mptcp_subflow_kern) !=
73
+    msk_owned_by_me(msk);
28
+         __alignof__(struct bpf_iter_mptcp_subflow));
74
+
29
75
+    kit->msk = msk;
30
    kit->msk = msk;
31
    if (!msk)
32
        return -EINVAL;
33
34
-    msk_owned_by_me(msk);
35
+    if (!sock_owned_by_user_nocheck(sk) &&
36
+     !spin_is_locked(&sk->sk_lock.slock))
37
+        return -EINVAL;
38
39
    kit->pos = &msk->conn_list;
76
    kit->pos = &msk->conn_list;
40
    return 0;
77
    return 0;
78
}
41
--
79
--
42
2.45.2
80
2.43.0
diff view generated by jsdifflib
[PATCH mptcp-next v5 3/5] Squash to "bpf: Register mptcp common kfunc set"
[PATCH mptcp-next v5 4/5] Revert "bpf: Acquire and release mptcp socket"
1
From: Geliang Tang <tanggeliang@kylinos.cn>
1
From: Geliang Tang <tanggeliang@kylinos.cn>
2
2
3
Drop bpf_mptcp_sk() and bpf_mptcp_subflow_tcp_sock() definitions. Use
3
Drop this patch as Martin suggested.
4
bpf_skc_to_mptcp_sock() and mptcp_subflow_tcp_sock() in mptcp_subflow
5
bpf_iter selftests instead.
6
4
7
Address Martin's comments in v1:
5
From Martin's review [1], this mptcp_sock_acquire() helper was a workaround
6
only to please the verifier, but they were not needed.
8
7
9
- add null-check for bpf_mptcp_subflow_ctx.
8
[1]
10
- add KF_RET_NULL flags for bpf_mptcp_subflow_ctx.
9
https://lore.kernel.org/9b373a23-c093-42d8-b4ae-99f2e62e7681@linux.dev
11
- register this kfunc set to BPF_PROG_TYPE_CGROUP_SOCKOPT only,
12
not BPF_PROG_TYPE_UNSPEC.
13
10
14
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
11
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
15
---
12
---
16
net/mptcp/bpf.c | 21 ++++++---------------
13
net/mptcp/bpf.c | 19 -------------------
17
1 file changed, 6 insertions(+), 15 deletions(-)
14
1 file changed, 19 deletions(-)
18
15
19
diff --git a/net/mptcp/bpf.c b/net/mptcp/bpf.c
16
diff --git a/net/mptcp/bpf.c b/net/mptcp/bpf.c
20
index XXXXXXX..XXXXXXX 100644
17
index XXXXXXX..XXXXXXX 100644
21
--- a/net/mptcp/bpf.c
18
--- a/net/mptcp/bpf.c
22
+++ b/net/mptcp/bpf.c
19
+++ b/net/mptcp/bpf.c
23
@@ -XXX,XX +XXX,XX @@ struct bpf_iter_mptcp_subflow_kern {
20
@@ -XXX,XX +XXX,XX @@ bpf_iter_mptcp_subflow_destroy(struct bpf_iter_mptcp_subflow *it)
24
21
{
25
__bpf_kfunc_start_defs();
22
}
26
23
27
-__bpf_kfunc static struct mptcp_sock *bpf_mptcp_sk(struct sock *sk)
24
-__bpf_kfunc static struct
25
-mptcp_sock *bpf_mptcp_sock_acquire(struct mptcp_sock *msk)
28
-{
26
-{
29
-    return mptcp_sk(sk);
27
-    struct sock *sk = (struct sock *)msk;
28
-
29
-    if (sk && refcount_inc_not_zero(&sk->sk_refcnt))
30
-        return msk;
31
-    return NULL;
30
-}
32
-}
31
-
33
-
32
__bpf_kfunc static struct mptcp_subflow_context *
34
-__bpf_kfunc static void bpf_mptcp_sock_release(struct mptcp_sock *msk)
33
bpf_mptcp_subflow_ctx(const struct sock *sk)
35
-{
36
-    struct sock *sk = (struct sock *)msk;
37
-
38
-    WARN_ON_ONCE(!sk || !refcount_dec_not_one(&sk->sk_refcnt));
39
-}
40
-
41
__bpf_kfunc struct mptcp_subflow_context *
42
bpf_mptcp_subflow_ctx_by_pos(const struct mptcp_sched_data *data, unsigned int pos)
34
{
43
{
35
-    return mptcp_subflow_ctx(sk);
44
@@ -XXX,XX +XXX,XX @@ BTF_ID_FLAGS(func, bpf_mptcp_subflow_ctx, KF_RET_NULL)
36
-}
37
+    if (sk && sk_fullsock(sk) &&
38
+     sk->sk_protocol == IPPROTO_TCP && sk_is_mptcp(sk))
39
+        return mptcp_subflow_ctx(sk);
40
41
-__bpf_kfunc static struct sock *
42
-bpf_mptcp_subflow_tcp_sock(const struct mptcp_subflow_context *subflow)
43
-{
44
-    return mptcp_subflow_tcp_sock(subflow);
45
+    return NULL;
46
}
47
48
__bpf_kfunc static int
49
@@ -XXX,XX +XXX,XX @@ __bpf_kfunc static bool bpf_mptcp_subflow_queues_empty(struct sock *sk)
50
__bpf_kfunc_end_defs();
51
52
BTF_KFUNCS_START(bpf_mptcp_common_kfunc_ids)
53
-BTF_ID_FLAGS(func, bpf_mptcp_sk)
54
-BTF_ID_FLAGS(func, bpf_mptcp_subflow_ctx)
55
-BTF_ID_FLAGS(func, bpf_mptcp_subflow_tcp_sock)
56
+BTF_ID_FLAGS(func, bpf_mptcp_subflow_ctx, KF_RET_NULL)
57
BTF_ID_FLAGS(func, bpf_iter_mptcp_subflow_new, KF_ITER_NEW | KF_TRUSTED_ARGS)
45
BTF_ID_FLAGS(func, bpf_iter_mptcp_subflow_new, KF_ITER_NEW | KF_TRUSTED_ARGS)
58
BTF_ID_FLAGS(func, bpf_iter_mptcp_subflow_next, KF_ITER_NEXT | KF_RET_NULL)
46
BTF_ID_FLAGS(func, bpf_iter_mptcp_subflow_next, KF_ITER_NEXT | KF_RET_NULL)
59
BTF_ID_FLAGS(func, bpf_iter_mptcp_subflow_destroy, KF_ITER_DESTROY)
47
BTF_ID_FLAGS(func, bpf_iter_mptcp_subflow_destroy, KF_ITER_DESTROY)
60
@@ -XXX,XX +XXX,XX @@ static int __init bpf_mptcp_kfunc_init(void)
48
-BTF_ID_FLAGS(func, bpf_mptcp_sock_acquire, KF_ACQUIRE | KF_RET_NULL)
61
    int ret;
49
-BTF_ID_FLAGS(func, bpf_mptcp_sock_release, KF_RELEASE)
62
50
BTF_KFUNCS_END(bpf_mptcp_common_kfunc_ids)
63
    ret = register_btf_fmodret_id_set(&bpf_mptcp_fmodret_set);
51
64
-    ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_UNSPEC,
52
static const struct btf_kfunc_id_set bpf_mptcp_common_kfunc_set = {
65
+    ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_CGROUP_SOCKOPT,
66
                     &bpf_mptcp_common_kfunc_set);
67
    ret = ret ?: register_btf_kfunc_id_set(BPF_PROG_TYPE_STRUCT_OPS,
68
                     &bpf_mptcp_sched_kfunc_set);
69
--
53
--
70
2.45.2
54
2.43.0
diff view generated by jsdifflib
[PATCH mptcp-next v5 5/5] Squash to "selftests/bpf: Add mptcp_subflow bpf_iter subtest"
[PATCH mptcp-next v5 5/5] Squash to "selftests/bpf: Add mptcp_subflow bpf_iter subtest"
1
From: Geliang Tang <tanggeliang@kylinos.cn>
1
From: Geliang Tang <tanggeliang@kylinos.cn>
2
2
3
Use bpf_skc_to_mptcp_sock() and mptcp_subflow_tcp_sock() instead of
3
Use bpf_core_cast() instead of bpf_skc_to_mptcp_sock().
4
bpf_mptcp_sk() and bpf_mptcp_subflow_tcp_sock().
4
Change the 2nd parameter type of bpf_for_each() as 'struct sock'.
5
Drop use of bpf_mptcp_sock_acquire/release.
6
Drop declaration of bpf_mptcp_subflow_tcp_sock. It's no longer used.
7
Update the comment for mptcp_subflow_tcp_sock(), which is a BPF helper,
8
not a kfunc.
5
9
6
IPPROTO_MPTCP is checked in bpf_skc_to_mptcp_sock(), no need to check
10
Please update the commit log as:
7
it in BPF program.
8
11
9
bpf_skc_to_mptcp_sock() and bpf_mptcp_subflow_ctx() may return NULL,
12
'''
10
need to check the return values.
13
This patch adds a "cgroup/getsockopt" program "iters_subflow" to test the
14
newly added mptcp_subflow bpf_iter.
15
16
Export mptcp_subflow helpers bpf_iter_mptcp_subflow_new/_next/_destroy
17
and other helpers into bpf_experimental.h.
18
19
Use bpf_for_each() to walk the subflow list of this msk. MPTCP-specific
20
packet scheduler kfunc can be called in the loop. In this test, just
21
add all subflow ids to local variable local_ids, then invoke the helper
22
mptcp_subflow_tcp_sock() in the loop to pick a subsocket.
23
24
Out of the loop, use bpf_mptcp_subflow_ctx() to get the subflow context
25
of the picked subsocket and do some verification. Finally, assign
26
local_ids to global variable ids so that the application can obtain this
27
value.
28
29
Add a subtest named test_iters_subflow to load and verify the newly added
30
mptcp_subflow type bpf_iter example in test_mptcp. Use the helper
31
endpoint_init() to add 3 new subflow endpoints. Send a byte of message
32
to start the mptcp connection, and wait for new subflows to be added.
33
getsockopt() is invoked to trigger the "cgroup/getsockopt" test program
34
"iters_subflow". Check if skel->bss->ids equals 10 to verify whether this
35
mptcp_subflow bpf_iter loops correctly as expected.
36
'''
11
37
12
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
38
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
13
---
39
---
14
tools/testing/selftests/bpf/progs/mptcp_bpf.h | 1 -
40
tools/testing/selftests/bpf/bpf_experimental.h | 2 +-
15
tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c | 11 +++++------
41
tools/testing/selftests/bpf/progs/mptcp_bpf.h | 5 -----
16
2 files changed, 5 insertions(+), 7 deletions(-)
42
tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c | 8 ++------
43
3 files changed, 3 insertions(+), 12 deletions(-)
17
44
45
diff --git a/tools/testing/selftests/bpf/bpf_experimental.h b/tools/testing/selftests/bpf/bpf_experimental.h
46
index XXXXXXX..XXXXXXX 100644
47
--- a/tools/testing/selftests/bpf/bpf_experimental.h
48
+++ b/tools/testing/selftests/bpf/bpf_experimental.h
49
@@ -XXX,XX +XXX,XX @@ extern void bpf_iter_css_destroy(struct bpf_iter_css *it) __weak __ksym;
50
51
struct bpf_iter_mptcp_subflow;
52
extern int bpf_iter_mptcp_subflow_new(struct bpf_iter_mptcp_subflow *it,
53
-                 struct mptcp_sock *msk) __weak __ksym;
54
+                 struct sock *sk) __weak __ksym;
55
extern struct mptcp_subflow_context *
56
bpf_iter_mptcp_subflow_next(struct bpf_iter_mptcp_subflow *it) __weak __ksym;
57
extern void
18
diff --git a/tools/testing/selftests/bpf/progs/mptcp_bpf.h b/tools/testing/selftests/bpf/progs/mptcp_bpf.h
58
diff --git a/tools/testing/selftests/bpf/progs/mptcp_bpf.h b/tools/testing/selftests/bpf/progs/mptcp_bpf.h
19
index XXXXXXX..XXXXXXX 100644
59
index XXXXXXX..XXXXXXX 100644
20
--- a/tools/testing/selftests/bpf/progs/mptcp_bpf.h
60
--- a/tools/testing/selftests/bpf/progs/mptcp_bpf.h
21
+++ b/tools/testing/selftests/bpf/progs/mptcp_bpf.h
61
+++ b/tools/testing/selftests/bpf/progs/mptcp_bpf.h
22
@@ -XXX,XX +XXX,XX @@ mptcp_subflow_tcp_sock(const struct mptcp_subflow_context *subflow)
62
@@ -XXX,XX +XXX,XX @@ mptcp_subflow_tcp_sock(const struct mptcp_subflow_context *subflow)
23
extern struct mptcp_sock *bpf_mptcp_sock_acquire(struct mptcp_sock *msk) __ksym;
63
}
24
extern void bpf_mptcp_sock_release(struct mptcp_sock *msk) __ksym;
64
25
65
/* ksym */
26
-extern struct mptcp_sock *bpf_mptcp_sk(struct sock *sk) __ksym;
66
-extern struct mptcp_sock *bpf_mptcp_sock_acquire(struct mptcp_sock *msk) __ksym;
67
-extern void bpf_mptcp_sock_release(struct mptcp_sock *msk) __ksym;
68
-
27
extern struct mptcp_subflow_context *
69
extern struct mptcp_subflow_context *
28
bpf_mptcp_subflow_ctx(const struct sock *sk) __ksym;
70
bpf_mptcp_subflow_ctx(const struct sock *sk) __ksym;
29
extern struct sock *
71
-extern struct sock *
72
-bpf_mptcp_subflow_tcp_sock(const struct mptcp_subflow_context *subflow) __ksym;
73
74
extern void mptcp_subflow_set_scheduled(struct mptcp_subflow_context *subflow,
75
                    bool scheduled) __ksym;
30
diff --git a/tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c b/tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c
76
diff --git a/tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c b/tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c
31
index XXXXXXX..XXXXXXX 100644
77
index XXXXXXX..XXXXXXX 100644
32
--- a/tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c
78
--- a/tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c
33
+++ b/tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c
79
+++ b/tools/testing/selftests/bpf/progs/mptcp_bpf_iters.c
34
@@ -XXX,XX +XXX,XX @@ int iters_subflow(struct bpf_sockopt *ctx)
80
@@ -XXX,XX +XXX,XX @@ int iters_subflow(struct bpf_sockopt *ctx)
35
    struct mptcp_sock *msk;
81
    if (!msk || msk->pm.server_side || !msk->pm.subflows)
36
    int local_ids = 0;
37
38
-    if (!sk || sk->protocol != IPPROTO_MPTCP ||
39
-     ctx->level != SOL_TCP || ctx->optname != TCP_IS_MPTCP)
40
+    if (ctx->level != SOL_TCP || ctx->optname != TCP_IS_MPTCP)
41
        return 1;
82
        return 1;
42
83
43
-    msk = bpf_mptcp_sk((struct sock *)sk);
84
-    msk = bpf_mptcp_sock_acquire(msk);
44
-    if (msk->pm.server_side || !msk->pm.subflows)
85
-    if (!msk)
45
+    msk = bpf_skc_to_mptcp_sock(sk);
86
-        return 1;
46
+    if (!msk || msk->pm.server_side || !msk->pm.subflows)
87
-    bpf_for_each(mptcp_subflow, subflow, msk) {
47
        return 1;
88
+    bpf_for_each(mptcp_subflow, subflow, (struct sock *)sk) {
48
89
        /* Here MPTCP-specific packet scheduler kfunc can be called:
49
    msk = bpf_mptcp_sock_acquire(msk);
90
         * this test is not doing anything really useful, only to
91
         * verify the iteration works.
50
@@ -XXX,XX +XXX,XX @@ int iters_subflow(struct bpf_sockopt *ctx)
92
@@ -XXX,XX +XXX,XX @@ int iters_subflow(struct bpf_sockopt *ctx)
93
51
        local_ids += subflow->subflow_id;
94
        local_ids += subflow->subflow_id;
52
95
53
        /* only to check the following kfunc works */
96
-        /* only to check the following kfunc works */
54
-        ssk = bpf_mptcp_subflow_tcp_sock(subflow);
97
+        /* only to check the following helper works */
55
+        ssk = mptcp_subflow_tcp_sock(subflow);
98
        ssk = mptcp_subflow_tcp_sock(subflow);
56
    }
99
    }
57
100
58
    if (!ssk)
59
@@ -XXX,XX +XXX,XX @@ int iters_subflow(struct bpf_sockopt *ctx)
101
@@ -XXX,XX +XXX,XX @@ int iters_subflow(struct bpf_sockopt *ctx)
60
61
    /* only to check the following kfunc works */
62
    subflow = bpf_mptcp_subflow_ctx(ssk);
63
-    if (subflow->token != msk->token)
64
+    if (!subflow || subflow->token != msk->token)
65
        goto out;
66
67
    ids = local_ids;
102
    ids = local_ids;
103
104
out:
105
-    bpf_mptcp_sock_release(msk);
106
    return 1;
107
}
68
--
108
--
69
2.45.2
109
2.43.0
diff view generated by jsdifflib

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.