v6:

- add "name" in struct mptcp_pm_ops.

- add some "sysctl" patches.

- drop "struct mptcp_pm_param".

- drop "pm_type" in mptcp_pm_data.

v5:

- use "struct mptcp_pm_param *param" as unified parameters for all

interfaces.

- register in-kernel mptcp_pm_ops too.

- only implement two interfaces "get_local_id" and "get_priority" in

this set.

v4:

- include a new patch "define BPF path manager type".

- add new interfaces:

    created established closed

    listerner_created listener_closed

- rename interfaces as:

    address_announced address_removed

    subflow_established subflow_closed

    get_priority set_priority

- rename functions as:

    mptcp_pm_validate

    mptcp_pm_register

    mptcp_pm_unregister

    mptcp_pm_initialize

    mptcp_pm_release

v3:

- rename the 2nd parameter of get_local_id() from 'local' to 'skc'.

- keep the 'msk_sport' check in mptcp_userspace_pm_get_local_id().

- return 'err' instead of '0' in userspace_pm_subflow_create().

- drop 'ret' variable inmptcp_pm_data_reset().

- fix typos in commit log.

v2:

- update get_local_id interface in patch 2.

From: Geliang Tang <tanggeliang@kylinos.cn>

The sysctl table needs to be updated when a new sysctl is added or an old

sysctl is deleted. It's inconvenient to use fixed index numbers, since the

index number that needs to be changed and all subsequent index numbers need

to be updated accordingly.

This patch adds an "i" variable to access each sysctl in the sysctl table

incrementally.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>

---

net/mptcp/ctrl.c | 23 ++++++++++++-----------

1 file changed, 12 insertions(+), 11 deletions(-)

diff --git a/net/mptcp/ctrl.c b/net/mptcp/ctrl.c

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/ctrl.c

+++ b/net/mptcp/ctrl.c

@@ -XXX,XX +XXX,XX @@ static int mptcp_pernet_new_table(struct net *net, struct mptcp_pernet *pernet)

{

    struct ctl_table_header *hdr;

    struct ctl_table *table;

+    int i = 0;

    table = mptcp_sysctl_table;

    if (!net_eq(net, &init_net)) {

@@ -XXX,XX +XXX,XX @@ static int mptcp_pernet_new_table(struct net *net, struct mptcp_pernet *pernet)

            goto err_alloc;

    }

-    table[0].data = &pernet->mptcp_enabled;

-    table[1].data = &pernet->add_addr_timeout;

-    table[2].data = &pernet->checksum_enabled;

-    table[3].data = &pernet->allow_join_initial_addr_port;

-    table[4].data = &pernet->stale_loss_cnt;

-    table[5].data = &pernet->pm_type;

-    table[6].data = &pernet->scheduler;

-    /* table[7] is for available_schedulers which is read-only info */

-    table[8].data = &pernet->close_timeout;

-    table[9].data = &pernet->blackhole_timeout;

-    table[10].data = &pernet->syn_retrans_before_tcp_fallback;

+    table[i++].data = &pernet->mptcp_enabled;

+    table[i++].data = &pernet->add_addr_timeout;

+    table[i++].data = &pernet->checksum_enabled;

+    table[i++].data = &pernet->allow_join_initial_addr_port;

+    table[i++].data = &pernet->stale_loss_cnt;

+    table[i++].data = &pernet->pm_type;

+    table[i++].data = &pernet->scheduler;

+    i++; /* table[i] is for available_schedulers which is read-only info */

+    table[i++].data = &pernet->close_timeout;

+    table[i++].data = &pernet->blackhole_timeout;

+    table[i++].data = &pernet->syn_retrans_before_tcp_fallback;

    hdr = register_net_sysctl_sz(net, MPTCP_SYSCTL_PATH, table,

                 ARRAY_SIZE(mptcp_sysctl_table));

--

2.43.0

From: Geliang Tang <tanggeliang@kylinos.cn>

A new net.mptcp.path_manager sysctl is added to determine which path

manager will be used by each newly-created MPTCP socket by setting the

name of it. This sysctl will replace the old one pm_type.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>

---

Documentation/networking/mptcp-sysctl.rst | 17 ++++++++

net/mptcp/ctrl.c | 50 +++++++++++++++++++++++

net/mptcp/protocol.h | 1 +

3 files changed, 68 insertions(+)

diff --git a/Documentation/networking/mptcp-sysctl.rst b/Documentation/networking/mptcp-sysctl.rst

index XXXXXXX..XXXXXXX 100644

--- a/Documentation/networking/mptcp-sysctl.rst

+++ b/Documentation/networking/mptcp-sysctl.rst

@@ -XXX,XX +XXX,XX @@ pm_type - INTEGER

    Default: 0

+path_manager - STRING

+    Set the default path manager name to use for each new MPTCP

+    socket. In-kernel path management will control subflow

+    connections and address advertisements according to

+    per-namespace values configured over the MPTCP netlink

+    API. Userspace path management puts per-MPTCP-connection subflow

+    connection decisions and address advertisements under control of

+    a privileged userspace program, at the cost of more netlink

+    traffic to propagate all of the related events and commands.

+

+    This is a per-namespace sysctl.

+

+    * "in-kernel" - In-kernel path manager

+    * "userspace" - Userspace path manager

+

+    Default: "in-kernel"

+

scheduler - STRING

    Select the scheduler of your choice.

diff --git a/net/mptcp/ctrl.c b/net/mptcp/ctrl.c

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/ctrl.c

+++ b/net/mptcp/ctrl.c

@@ -XXX,XX +XXX,XX @@ struct mptcp_pernet {

    u8 checksum_enabled;

    u8 allow_join_initial_addr_port;

    u8 pm_type;

+    char path_manager[MPTCP_PM_NAME_MAX];

    char scheduler[MPTCP_SCHED_NAME_MAX];

};

@@ -XXX,XX +XXX,XX @@ int mptcp_get_pm_type(const struct net *net)

    return mptcp_get_pernet(net)->pm_type;

}

+const char *mptcp_get_path_manager(const struct net *net)

+{

+    return mptcp_get_pernet(net)->path_manager;

+}

+

const char *mptcp_get_scheduler(const struct net *net)

{

    return mptcp_get_pernet(net)->scheduler;

@@ -XXX,XX +XXX,XX @@ static void mptcp_pernet_set_defaults(struct mptcp_pernet *pernet)

    pernet->allow_join_initial_addr_port = 1;

    pernet->stale_loss_cnt = 4;

    pernet->pm_type = MPTCP_PM_TYPE_KERNEL;

+    strscpy(pernet->path_manager, "in-kernel", sizeof(pernet->path_manager));

    strscpy(pernet->scheduler, "default", sizeof(pernet->scheduler));

}

#ifdef CONFIG_SYSCTL

+static int mptcp_set_path_manager(char *path_manager, const char *name)

+{

+    struct mptcp_pm_ops *pm;

+    int ret = 0;

+

+    rcu_read_lock();

+    pm = mptcp_pm_find(name);

+    if (pm)

+        strscpy(path_manager, name, MPTCP_PM_NAME_MAX);

+    else

+        ret = -ENOENT;

+    rcu_read_unlock();

+

+    return ret;

+}

+

+static int proc_path_manager(const struct ctl_table *ctl, int write,

+             void *buffer, size_t *lenp, loff_t *ppos)

+{

+    char (*path_manager)[MPTCP_PM_NAME_MAX] = ctl->data;

+    char val[MPTCP_PM_NAME_MAX];

+    const struct ctl_table tbl = {

+        .data = val,

+        .maxlen = MPTCP_PM_NAME_MAX,

+    };

+    int ret;

+

+    strscpy(val, *path_manager, MPTCP_PM_NAME_MAX);

+

+    ret = proc_dostring(&tbl, write, buffer, lenp, ppos);

+    if (write && ret == 0)

+        ret = mptcp_set_path_manager(*path_manager, val);

+

+    return ret;

+}

+

static int mptcp_set_scheduler(char *scheduler, const char *name)

{

    struct mptcp_sched_ops *sched;

@@ -XXX,XX +XXX,XX @@ static struct ctl_table mptcp_sysctl_table[] = {

        .extra1 = SYSCTL_ZERO,

        .extra2 = &mptcp_pm_type_max

    },

+    {

+        .procname = "path_manager",

+        .maxlen    = MPTCP_PM_NAME_MAX,

+        .mode = 0644,

+        .proc_handler = proc_path_manager,

+    },

    {

        .procname = "scheduler",

        .maxlen    = MPTCP_SCHED_NAME_MAX,

@@ -XXX,XX +XXX,XX @@ static int mptcp_pernet_new_table(struct net *net, struct mptcp_pernet *pernet)

    table[i++].data = &pernet->allow_join_initial_addr_port;

    table[i++].data = &pernet->stale_loss_cnt;

    table[i++].data = &pernet->pm_type;

+    table[i++].data = &pernet->path_manager;

    table[i++].data = &pernet->scheduler;

    i++; /* table[i] is for available_schedulers which is read-only info */

    table[i++].data = &pernet->close_timeout;

diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/protocol.h

+++ b/net/mptcp/protocol.h

@@ -XXX,XX +XXX,XX @@ int mptcp_allow_join_id0(const struct net *net);

unsigned int mptcp_stale_loss_cnt(const struct net *net);

unsigned int mptcp_close_timeout(const struct sock *sk);

int mptcp_get_pm_type(const struct net *net);

+const char *mptcp_get_path_manager(const struct net *net);

const char *mptcp_get_scheduler(const struct net *net);

void mptcp_active_disable(struct sock *sk);

--

2.43.0

From: Geliang Tang <tanggeliang@kylinos.cn>

The following code in mptcp_userspace_pm_get_local_id() that assigns "skc"

to "new_entry" is not allowed in BPF if we use the same code to implement

the get_local_id() interface of a BFP path manager:

    memset(&new_entry, 0, sizeof(struct mptcp_pm_addr_entry));

    new_entry.addr = *skc;

    new_entry.addr.id = 0;

    new_entry.flags = MPTCP_PM_ADDR_FLAG_IMPLICIT;

To solve the issue, this patch moves this assignment to "new_entry" forward

to mptcp_pm_get_local_id(), and then passing "new_entry" as a parameter to

both mptcp_pm_nl_get_local_id() and mptcp_userspace_pm_get_local_id().

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>

---

net/mptcp/pm.c | 9 ++++++---

net/mptcp/pm_netlink.c | 11 ++++-------

net/mptcp/pm_userspace.c | 17 ++++++-----------

net/mptcp/protocol.h | 6 ++++--

4 files changed, 20 insertions(+), 23 deletions(-)

diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/pm.c

+++ b/net/mptcp/pm.c

@@ -XXX,XX +XXX,XX @@ bool mptcp_pm_rm_addr_signal(struct mptcp_sock *msk, unsigned int remaining,

int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc)

{

-    struct mptcp_addr_info skc_local;

+    struct mptcp_pm_addr_entry skc_local = { 0 };

    struct mptcp_addr_info msk_local;

    if (WARN_ON_ONCE(!msk))

@@ -XXX,XX +XXX,XX @@ int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc)

     * addr

     */

    mptcp_local_address((struct sock_common *)msk, &msk_local);

-    mptcp_local_address((struct sock_common *)skc, &skc_local);

-    if (mptcp_addresses_equal(&msk_local, &skc_local, false))

+    mptcp_local_address((struct sock_common *)skc, &skc_local.addr);

+    if (mptcp_addresses_equal(&msk_local, &skc_local.addr, false))

        return 0;

+    skc_local.addr.id = 0;

+    skc_local.flags = MPTCP_PM_ADDR_FLAG_IMPLICIT;

+

    if (mptcp_pm_is_userspace(msk))

        return mptcp_userspace_pm_get_local_id(msk, &skc_local);

    return mptcp_pm_nl_get_local_id(msk, &skc_local);

diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/pm_netlink.c

+++ b/net/mptcp/pm_netlink.c

@@ -XXX,XX +XXX,XX @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk,

    return err;

}

-int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc)

+int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk,

+             struct mptcp_pm_addr_entry *skc)

{

    struct mptcp_pm_addr_entry *entry;

    struct pm_nl_pernet *pernet;

@@ -XXX,XX +XXX,XX @@ int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc

    pernet = pm_nl_get_pernet_from_msk(msk);

    rcu_read_lock();

-    entry = __lookup_addr(pernet, skc);

+    entry = __lookup_addr(pernet, &skc->addr);

    ret = entry ? entry->addr.id : -1;

    rcu_read_unlock();

    if (ret >= 0)

@@ -XXX,XX +XXX,XX @@ int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc

    if (!entry)

        return -ENOMEM;

-    entry->addr = *skc;

-    entry->addr.id = 0;

+    *entry = *skc;

    entry->addr.port = 0;

-    entry->ifindex = 0;

-    entry->flags = MPTCP_PM_ADDR_FLAG_IMPLICIT;

-    entry->lsk = NULL;

    ret = mptcp_pm_nl_append_new_local_addr(pernet, entry, true);

    if (ret < 0)

        kfree(entry);

diff --git a/net/mptcp/pm_userspace.c b/net/mptcp/pm_userspace.c

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/pm_userspace.c

+++ b/net/mptcp/pm_userspace.c

@@ -XXX,XX +XXX,XX @@ mptcp_userspace_pm_lookup_addr_by_id(struct mptcp_sock *msk, unsigned int id)

}

int mptcp_userspace_pm_get_local_id(struct mptcp_sock *msk,

-                 struct mptcp_addr_info *skc)

+                 struct mptcp_pm_addr_entry *skc)

{

-    struct mptcp_pm_addr_entry *entry = NULL, new_entry;

    __be16 msk_sport = ((struct inet_sock *)

             inet_sk((struct sock *)msk))->inet_sport;

+    struct mptcp_pm_addr_entry *entry;

    spin_lock_bh(&msk->pm.lock);

-    entry = mptcp_userspace_pm_lookup_addr(msk, skc);

+    entry = mptcp_userspace_pm_lookup_addr(msk, &skc->addr);

    spin_unlock_bh(&msk->pm.lock);

    if (entry)

        return entry->addr.id;

-    memset(&new_entry, 0, sizeof(struct mptcp_pm_addr_entry));

-    new_entry.addr = *skc;

-    new_entry.addr.id = 0;

-    new_entry.flags = MPTCP_PM_ADDR_FLAG_IMPLICIT;

-

-    if (new_entry.addr.port == msk_sport)

-        new_entry.addr.port = 0;

+    if (skc->addr.port == msk_sport)

+        skc->addr.port = 0;

-    return mptcp_userspace_pm_append_new_local_addr(msk, &new_entry, true);

+    return mptcp_userspace_pm_append_new_local_addr(msk, skc, true);

}

bool mptcp_userspace_pm_is_backup(struct mptcp_sock *msk,

diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/protocol.h

+++ b/net/mptcp/protocol.h

@@ -XXX,XX +XXX,XX @@ bool mptcp_pm_add_addr_signal(struct mptcp_sock *msk, const struct sk_buff *skb,

bool mptcp_pm_rm_addr_signal(struct mptcp_sock *msk, unsigned int remaining,

             struct mptcp_rm_list *rm_list);

int mptcp_pm_get_local_id(struct mptcp_sock *msk, struct sock_common *skc);

-int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc);

-int mptcp_userspace_pm_get_local_id(struct mptcp_sock *msk, struct mptcp_addr_info *skc);

+int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk,

+             struct mptcp_pm_addr_entry *skc);

+int mptcp_userspace_pm_get_local_id(struct mptcp_sock *msk,

+                 struct mptcp_pm_addr_entry *skc);

bool mptcp_pm_is_backup(struct mptcp_sock *msk, struct sock_common *skc);

bool mptcp_pm_nl_is_backup(struct mptcp_sock *msk, struct mptcp_addr_info *skc);

bool mptcp_userspace_pm_is_backup(struct mptcp_sock *msk, struct mptcp_addr_info *skc);

--

2.43.0

+#define MPTCP_PM_NAME_MAX    16

+

+    char            name[MPTCP_PM_NAME_MAX];

#include <net/mptcp.h>

From: Geliang Tang <tanggeliang@kylinos.cn>

This patch defines the original in-kernel netlink path manager as a new

struct mptcp_pm_ops named "mptcp_in_kernel_pm", and register it in

mptcp_pm_nl_init().

This mptcp_pm_ops will be skipped in mptcp_pm_unregister().

Only get_local_id() and get_priority() interfaces are implemented here.

mptcp_pm_nl_is_backup() becomes a wrapper of get_priority().

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>

---

net/mptcp/pm.c | 3 +++

net/mptcp/pm_netlink.c | 18 +++++++++++++++++-

net/mptcp/protocol.h | 2 ++

3 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/pm.c

+++ b/net/mptcp/pm.c

@@ -XXX,XX +XXX,XX @@ int mptcp_pm_register(struct mptcp_pm_ops *pm)

void mptcp_pm_unregister(struct mptcp_pm_ops *pm)

{

+    if (pm == &mptcp_in_kernel_pm)

+        return;

+

    spin_lock(&mptcp_pm_list_lock);

    list_del_rcu(&pm->list);

    spin_unlock(&mptcp_pm_list_lock);

diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/pm_netlink.c

+++ b/net/mptcp/pm_netlink.c

@@ -XXX,XX +XXX,XX @@ int mptcp_pm_nl_get_local_id(struct mptcp_sock *msk,

    return ret;

}

-bool mptcp_pm_nl_is_backup(struct mptcp_sock *msk, struct mptcp_addr_info *skc)

+static bool mptcp_pm_nl_get_priority(struct mptcp_sock *msk,

+                 struct mptcp_addr_info *skc)

{

    struct pm_nl_pernet *pernet = pm_nl_get_pernet_from_msk(msk);

    struct mptcp_pm_addr_entry *entry;

@@ -XXX,XX +XXX,XX @@ bool mptcp_pm_nl_is_backup(struct mptcp_sock *msk, struct mptcp_addr_info *skc)

    return backup;

}

+bool mptcp_pm_nl_is_backup(struct mptcp_sock *msk, struct mptcp_addr_info *skc)

+{

+    return mptcp_pm_nl_get_priority(msk, skc);

+}

+

#define MPTCP_PM_CMD_GRP_OFFSET 0

#define MPTCP_PM_EV_GRP_OFFSET 1

@@ -XXX,XX +XXX,XX @@ static struct pernet_operations mptcp_pm_pernet_ops = {

    .size = sizeof(struct pm_nl_pernet),

};

+struct mptcp_pm_ops mptcp_in_kernel_pm = {

+    .get_local_id        = mptcp_pm_nl_get_local_id,

+    .get_priority        = mptcp_pm_nl_get_priority,

+    .type            = MPTCP_PM_TYPE_KERNEL,

+    .name            = "in-kernel",

+    .owner            = THIS_MODULE,

+};

+

void __init mptcp_pm_nl_init(void)

{

    if (register_pernet_subsys(&mptcp_pm_pernet_ops) < 0)

@@ -XXX,XX +XXX,XX @@ void __init mptcp_pm_nl_init(void)

    if (genl_register_family(&mptcp_genl_family))

        panic("Failed to register MPTCP PM netlink family\n");

+

+    mptcp_pm_register(&mptcp_in_kernel_pm);

}

diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h

index XXXXXXX..XXXXXXX 100644

--- a/net/mptcp/protocol.h

+++ b/net/mptcp/protocol.h

@@ -XXX,XX +XXX,XX @@ int mptcp_pm_remove_addr(struct mptcp_sock *msk, const struct mptcp_rm_list *rm_

void mptcp_pm_remove_addr_entry(struct mptcp_sock *msk,

                struct mptcp_pm_addr_entry *entry);

+extern struct mptcp_pm_ops mptcp_in_kernel_pm;

+

struct mptcp_pm_ops *mptcp_pm_find(const char *name);

int mptcp_pm_validate(struct mptcp_pm_ops *pm);

int mptcp_pm_register(struct mptcp_pm_ops *pm);

--

2.43.0

/* Must be called with rcu read lock held */

+    .name            = "userspace",

+    const char *path_manager = mptcp_get_path_manager(sock_net((struct sock *)msk));