[PATCH mptcp-net 0/3] fixes for userspace pm

[PATCH mptcp-net 0/3] fixes for userspace pm

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

Three fixes for userspace pm extract from "BPF path manager" series.

Geliang Tang (3):
  mptcp: add lookup_addr for userspace pm
  mptcp: update local address flags when setting it
  mptcp: hold pm lock when deleting entry

 net/mptcp/pm_userspace.c | 72 ++++++++++++++++++++++++----------------
 1 file changed, 44 insertions(+), 28 deletions(-)

-- 
2.45.2

[PATCH mptcp-net 1/3] mptcp: add lookup_addr for userspace pm

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

Like __lookup_addr() helper in pm_netlink.c, a new helper
mptcp_userspace_pm_lookup_addr() is also defined in pm_userspace.c.
It looks up the corresponding mptcp_pm_addr_entry address in
userspace_pm_local_addr_list through the passed "addr" parameter
and returns it.

This helper can be used in mptcp_userspace_pm_delete_local_addr(),
mptcp_userspace_pm_get_local_id() and mptcp_userspace_pm_is_backup()
to simplify the code.

Please note that with this change now list_for_each_entry() is used in
mptcp_userspace_pm_append_new_local_addr(), not list_for_each_entry_safe(),
but that's OK to do so because mptcp_userspace_pm_lookup_addr() only
returns an entry from the list, the list hasn't been modified here.

Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/pm_userspace.c | 58 +++++++++++++++++++++-------------------
 1 file changed, 30 insertions(+), 28 deletions(-)

diff --git a/net/mptcp/pm_userspace.c b/net/mptcp/pm_userspace.c
index 56dfea9862b7..9c622b0e3e6e 100644
--- a/net/mptcp/pm_userspace.c
+++ b/net/mptcp/pm_userspace.c
@@ -26,6 +26,19 @@ void mptcp_free_local_addr_list(struct mptcp_sock *msk)
 	}
 }
 
+static struct mptcp_pm_addr_entry *
+mptcp_userspace_pm_lookup_addr(struct mptcp_sock *msk,
+			       const struct mptcp_addr_info *addr)
+{
+	struct mptcp_pm_addr_entry *entry;
+
+	list_for_each_entry(entry, &msk->pm.userspace_pm_local_addr_list, list) {
+		if (mptcp_addresses_equal(&entry->addr, addr, false))
+			return entry;
+	}
+	return NULL;
+}
+
 static int mptcp_userspace_pm_append_new_local_addr(struct mptcp_sock *msk,
 						    struct mptcp_pm_addr_entry *entry,
 						    bool needs_id)
@@ -90,22 +103,20 @@ static int mptcp_userspace_pm_append_new_local_addr(struct mptcp_sock *msk,
 static int mptcp_userspace_pm_delete_local_addr(struct mptcp_sock *msk,
 						struct mptcp_pm_addr_entry *addr)
 {
-	struct mptcp_pm_addr_entry *entry, *tmp;
 	struct sock *sk = (struct sock *)msk;
+	struct mptcp_pm_addr_entry *entry;
 
-	list_for_each_entry_safe(entry, tmp, &msk->pm.userspace_pm_local_addr_list, list) {
-		if (mptcp_addresses_equal(&entry->addr, &addr->addr, false)) {
-			/* TODO: a refcount is needed because the entry can
-			 * be used multiple times (e.g. fullmesh mode).
-			 */
-			list_del_rcu(&entry->list);
-			sock_kfree_s(sk, entry, sizeof(*entry));
-			msk->pm.local_addr_used--;
-			return 0;
-		}
-	}
-
-	return -EINVAL;
+	entry = mptcp_userspace_pm_lookup_addr(msk, &addr->addr);
+	if (!entry)
+		return -EINVAL;
+
+	/* TODO: a refcount is needed because the entry can
+	 * be used multiple times (e.g. fullmesh mode).
+	 */
+	list_del_rcu(&entry->list);
+	sock_kfree_s(sk, entry, sizeof(*entry));
+	msk->pm.local_addr_used--;
+	return 0;
 }
 
 static struct mptcp_pm_addr_entry *
@@ -123,17 +134,12 @@ 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 *entry = NULL, *e, new_entry;
+	struct mptcp_pm_addr_entry *entry = NULL, new_entry;
 	__be16 msk_sport =  ((struct inet_sock *)
 			     inet_sk((struct sock *)msk))->inet_sport;
 
 	spin_lock_bh(&msk->pm.lock);
-	list_for_each_entry(e, &msk->pm.userspace_pm_local_addr_list, list) {
-		if (mptcp_addresses_equal(&e->addr, skc, false)) {
-			entry = e;
-			break;
-		}
-	}
+	entry = mptcp_userspace_pm_lookup_addr(msk, skc);
 	spin_unlock_bh(&msk->pm.lock);
 	if (entry)
 		return entry->addr.id;
@@ -153,15 +159,11 @@ bool mptcp_userspace_pm_is_backup(struct mptcp_sock *msk,
 				  struct mptcp_addr_info *skc)
 {
 	struct mptcp_pm_addr_entry *entry;
-	bool backup = false;
+	bool backup;
 
 	spin_lock_bh(&msk->pm.lock);
-	list_for_each_entry(entry, &msk->pm.userspace_pm_local_addr_list, list) {
-		if (mptcp_addresses_equal(&entry->addr, skc, false)) {
-			backup = !!(entry->flags & MPTCP_PM_ADDR_FLAG_BACKUP);
-			break;
-		}
-	}
+	entry = mptcp_userspace_pm_lookup_addr(msk, skc);
+	backup = entry && !!(entry->flags & MPTCP_PM_ADDR_FLAG_BACKUP);
 	spin_unlock_bh(&msk->pm.lock);
 
 	return backup;
-- 
2.45.2

Re: [PATCH mptcp-net 1/3] mptcp: add lookup_addr for userspace pm

[Matthieu Baerts]

Hi Geliang,

Thank you for the patches!

On 05/11/2024 10:40, Geliang Tang wrote:
> From: Geliang Tang <tanggeliang@kylinos.cn>
> 
> Like __lookup_addr() helper in pm_netlink.c, a new helper
> mptcp_userspace_pm_lookup_addr() is also defined in pm_userspace.c.
> It looks up the corresponding mptcp_pm_addr_entry address in
> userspace_pm_local_addr_list through the passed "addr" parameter
> and returns it.
> 
> This helper can be used in mptcp_userspace_pm_delete_local_addr(),
> mptcp_userspace_pm_get_local_id() and mptcp_userspace_pm_is_backup()
> to simplify the code.

I would prefer to avoid the refactoring for -net if it is not needed:
this will make the next fix harder to backport. Please see my comment on
the next patch.

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.

[PATCH mptcp-net 2/3] mptcp: update local address flags when setting it

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

Just like in-kernel pm, when userspace pm does set_flags, it needs to send
out MP_PRIO signal, and also modify the flags of the corresponding address
entry in the local address list. This patch implements the missing logic.

Use mptcp_userspace_pm_lookup_addr() helper to find the address entry on
userspace_pm_local_addr_list, if bkup is true, set the flags of the address
entry with FLAG_BACKUP, otherwise, clear FLAG_BACKUP.

Fixes: 892f396c8e68 ("mptcp: netlink: issue MP_PRIO signals from userspace PMs")
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/pm_userspace.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/net/mptcp/pm_userspace.c b/net/mptcp/pm_userspace.c
index 9c622b0e3e6e..0c0f6f65accb 100644
--- a/net/mptcp/pm_userspace.c
+++ b/net/mptcp/pm_userspace.c
@@ -562,6 +562,7 @@ int mptcp_userspace_pm_set_flags(struct sk_buff *skb, struct genl_info *info)
 	struct nlattr *token = info->attrs[MPTCP_PM_ATTR_TOKEN];
 	struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR];
 	struct net *net = sock_net(skb->sk);
+	struct mptcp_pm_addr_entry *entry;
 	struct mptcp_sock *msk;
 	int ret = -EINVAL;
 	struct sock *sk;
@@ -603,6 +604,16 @@ int mptcp_userspace_pm_set_flags(struct sk_buff *skb, struct genl_info *info)
 	if (loc.flags & MPTCP_PM_ADDR_FLAG_BACKUP)
 		bkup = 1;
 
+	spin_lock_bh(&msk->pm.lock);
+	entry = mptcp_userspace_pm_lookup_addr(msk, &loc.addr);
+	if (entry) {
+		if (bkup)
+			entry->flags |= MPTCP_PM_ADDR_FLAG_BACKUP;
+		else
+			entry->flags &= ~MPTCP_PM_ADDR_FLAG_BACKUP;
+	}
+	spin_unlock_bh(&msk->pm.lock);
+
 	lock_sock(sk);
 	ret = mptcp_pm_nl_mp_prio_send_ack(msk, &loc.addr, &rem.addr, bkup);
 	release_sock(sk);
-- 
2.45.2

Re: [PATCH mptcp-net 2/3] mptcp: update local address flags when setting it

[Matthieu Baerts]

Hi Geliang,

On 05/11/2024 10:40, Geliang Tang wrote:
> From: Geliang Tang <tanggeliang@kylinos.cn>
> 
> Just like in-kernel pm, when userspace pm does set_flags, it needs to send
> out MP_PRIO signal, and also modify the flags of the corresponding address
> entry in the local address list. This patch implements the missing logic.
> 
> Use mptcp_userspace_pm_lookup_addr() helper to find the address entry on
> userspace_pm_local_addr_list, if bkup is true, set the flags of the address
> entry with FLAG_BACKUP, otherwise, clear FLAG_BACKUP.
> 
> Fixes: 892f396c8e68 ("mptcp: netlink: issue MP_PRIO signals from userspace PMs")
> Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
> ---
>  net/mptcp/pm_userspace.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)
> 
> diff --git a/net/mptcp/pm_userspace.c b/net/mptcp/pm_userspace.c
> index 9c622b0e3e6e..0c0f6f65accb 100644
> --- a/net/mptcp/pm_userspace.c
> +++ b/net/mptcp/pm_userspace.c
> @@ -562,6 +562,7 @@ int mptcp_userspace_pm_set_flags(struct sk_buff *skb, struct genl_info *info)
>  	struct nlattr *token = info->attrs[MPTCP_PM_ATTR_TOKEN];
>  	struct nlattr *attr = info->attrs[MPTCP_PM_ATTR_ADDR];
>  	struct net *net = sock_net(skb->sk);
> +	struct mptcp_pm_addr_entry *entry;
>  	struct mptcp_sock *msk;
>  	int ret = -EINVAL;
>  	struct sock *sk;
> @@ -603,6 +604,16 @@ int mptcp_userspace_pm_set_flags(struct sk_buff *skb, struct genl_info *info)
>  	if (loc.flags & MPTCP_PM_ADDR_FLAG_BACKUP)
>  		bkup = 1;
>  
> +	spin_lock_bh(&msk->pm.lock);
> +	entry = mptcp_userspace_pm_lookup_addr(msk, &loc.addr);

(Linked to my comment from the previous patch) Adding only the helper,
and squashing that in this patch here might be an option, but even that
can cause conflicts.

I think the simplest is not to add this small helper only used once for
the moment: so simply calling list_for_each_entry() here, and add the
following code inside it. WDYT?

> +	if (entry) {
> +		if (bkup)
> +			entry->flags |= MPTCP_PM_ADDR_FLAG_BACKUP;
> +		else
> +			entry->flags &= ~MPTCP_PM_ADDR_FLAG_BACKUP;
> +	}
> +	spin_unlock_bh(&msk->pm.lock);
> +
>  	lock_sock(sk);
>  	ret = mptcp_pm_nl_mp_prio_send_ack(msk, &loc.addr, &rem.addr, bkup);
>  	release_sock(sk);

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.

[PATCH mptcp-net 3/3] mptcp: hold pm lock when deleting entry

[Geliang Tang]

From: Geliang Tang <tanggeliang@kylinos.cn>

When traversing userspace_pm_local_addr_list and deleting an entry from
it in mptcp_pm_nl_remove_doit(), msk->pm.lock should be held. This patch
holds the lock in mptcp_pm_nl_remove_doit().

Fixes: d9a4594edabf ("mptcp: netlink: Add MPTCP_PM_CMD_REMOVE")
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn>
---
 net/mptcp/pm_userspace.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/mptcp/pm_userspace.c b/net/mptcp/pm_userspace.c
index 0c0f6f65accb..3664f3c1572e 100644
--- a/net/mptcp/pm_userspace.c
+++ b/net/mptcp/pm_userspace.c
@@ -310,14 +310,17 @@ int mptcp_pm_nl_remove_doit(struct sk_buff *skb, struct genl_info *info)
 
 	lock_sock(sk);
 
+	spin_lock_bh(&msk->pm.lock);
 	match = mptcp_userspace_pm_lookup_addr_by_id(msk, id_val);
 	if (!match) {
 		GENL_SET_ERR_MSG(info, "address with specified id not found");
+		spin_unlock_bh(&msk->pm.lock);
 		release_sock(sk);
 		goto out;
 	}
 
 	list_move(&match->list, &free_list);
+	spin_unlock_bh(&msk->pm.lock);
 
 	mptcp_pm_remove_addrs(msk, &free_list);
 
-- 
2.45.2

Re: [PATCH mptcp-net 0/3] fixes for userspace pm

[MPTCP CI]

Hi Geliang,

Thank you for your modifications, that's great!

Our CI did some validations and here is its report:

- KVM Validation: normal: Critical: Global Timeout ❌
- KVM Validation: debug: Success! ✅
- KVM Validation: btf-normal (only bpftest_all): Success! ✅
- KVM Validation: btf-debug (only bpftest_all): Success! ✅
- Task: https://github.com/multipath-tcp/mptcp_net-next/actions/runs/11681780257

Initiator: Patchew Applier
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/e594c42ba8aa
Patchwork: https://patchwork.kernel.org/project/mptcp/list/?series=906422


If there are some issues, you can reproduce them using the same environment as
the one used by the CI thanks to a docker image, e.g.:

    $ cd [kernel source code]
    $ docker run -v "${PWD}:${PWD}:rw" -w "${PWD}" --privileged --rm -it \
        --pull always mptcp/mptcp-upstream-virtme-docker:latest \
        auto-normal

For more details:

    https://github.com/multipath-tcp/mptcp-upstream-virtme-docker


Please note that despite all the efforts that have been already done to have a
stable tests suite when executed on a public CI like here, it is possible some
reported issues are not due to your modifications. Still, do not hesitate to
help us improve that ;-)

Cheers,
MPTCP GH Action bot
Bot operated by Matthieu Baerts (NGI0 Core)

Re: [PATCH mptcp-net 0/3] fixes for userspace pm

[MPTCP CI]

Hi Geliang,

Thank you for your modifications, that's great!

Our CI did some validations and here is its report:

- KVM Validation: normal: Success! ✅
- KVM Validation: debug: Success! ✅
- KVM Validation: btf-normal (only bpftest_all): Success! ✅
- KVM Validation: btf-debug (only bpftest_all): Success! ✅
- Task: https://github.com/multipath-tcp/mptcp_net-next/actions/runs/11681780257

Initiator: Patchew Applier
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/e594c42ba8aa
Patchwork: https://patchwork.kernel.org/project/mptcp/list/?series=906422


If there are some issues, you can reproduce them using the same environment as
the one used by the CI thanks to a docker image, e.g.:

    $ cd [kernel source code]
    $ docker run -v "${PWD}:${PWD}:rw" -w "${PWD}" --privileged --rm -it \
        --pull always mptcp/mptcp-upstream-virtme-docker:latest \
        auto-normal

For more details:

    https://github.com/multipath-tcp/mptcp-upstream-virtme-docker


Please note that despite all the efforts that have been already done to have a
stable tests suite when executed on a public CI like here, it is possible some
reported issues are not due to your modifications. Still, do not hesitate to
help us improve that ;-)

Cheers,
MPTCP GH Action bot
Bot operated by Matthieu Baerts (NGI0 Core)