[PATCH v2 mptcp-next 1/3] mptcp: optimize release_cb for the common case

[PATCH v2 mptcp-next 1/3] mptcp: optimize release_cb for the common case

[Paolo Abeni]

The mptcp release callback checks several flags in atomic
context, but only MPTCP_CLEAN_UNA can be up frequently.

Reorganize the code to avoid multiple conditionals in the
most common scenarios.

Additional clarify a related comment.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
---
 net/mptcp/protocol.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index fbb14dfe62b3..bdba1ddee2a7 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -3118,15 +3118,17 @@ static void mptcp_release_cb(struct sock *sk)
 		spin_lock_bh(&sk->sk_lock.slock);
 	}
 
-	/* be sure to set the current sk state before tacking actions
-	 * depending on sk_state
-	 */
-	if (__test_and_clear_bit(MPTCP_CONNECTED, &msk->cb_flags))
-		__mptcp_set_connected(sk);
 	if (__test_and_clear_bit(MPTCP_CLEAN_UNA, &msk->cb_flags))
 		__mptcp_clean_una_wakeup(sk);
-	if (__test_and_clear_bit(MPTCP_ERROR_REPORT, &msk->cb_flags))
-		__mptcp_error_report(sk);
+	if (unlikely(&msk->cb_flags)) {
+		/* be sure to set the current sk state before tacking actions
+		 * depending on sk_state, that is processing MPTCP_ERROR_REPORT
+		 */
+		if (__test_and_clear_bit(MPTCP_CONNECTED, &msk->cb_flags))
+			__mptcp_set_connected(sk);
+		if (__test_and_clear_bit(MPTCP_ERROR_REPORT, &msk->cb_flags))
+			__mptcp_error_report(sk);
+	}
 
 	__mptcp_update_rmem(sk);
 }
-- 
2.35.1

[PATCH v2 mptcp-next 2/3] mptcp: reset the packet scheduler on incoming MP_PRIO.

[Paolo Abeni]

When an incoming MP_PRIO option changes the backup
status of any subflow, we need to reset the packet
scheduler status, or the next send could keep using
the previously selected subflow, without taking in account
the new priorities.

Reported-by: Davide Caratti <dcaratti@redhat.com>
Fixes: 40453a5c61f4 ("mptcp: add the incoming MP_PRIO support")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
v1 -> v2:
 - fix backup comparison - Mat
---
 net/mptcp/pm.c       | 19 +++++++++++++++----
 net/mptcp/protocol.c |  2 ++
 net/mptcp/protocol.h |  1 +
 3 files changed, 18 insertions(+), 4 deletions(-)

diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c
index d0d31d5c198a..c87f8974a1a7 100644
--- a/net/mptcp/pm.c
+++ b/net/mptcp/pm.c
@@ -262,14 +262,25 @@ void mptcp_pm_rm_addr_received(struct mptcp_sock *msk,
 	spin_unlock_bh(&pm->lock);
 }
 
-void mptcp_pm_mp_prio_received(struct sock *sk, u8 bkup)
+void mptcp_pm_mp_prio_received(struct sock *ssk, u8 bkup)
 {
-	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
+	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk);
+	struct sock *sk = subflow->conn;
+	struct mptcp_sock *msk;
 
 	pr_debug("subflow->backup=%d, bkup=%d\n", subflow->backup, bkup);
-	subflow->backup = bkup;
+	msk = mptcp_sk(sk);
+	if (subflow->backup != bkup) {
+		subflow->backup = bkup;
+		mptcp_data_lock(sk);
+		if (!sock_owned_by_user(sk))
+			msk->last_snd = NULL;
+		else
+			__set_bit(MPTCP_RESET_SCHEDULER,  &msk->cb_flags);
+		mptcp_data_unlock(sk);
+	}
 
-	mptcp_event(MPTCP_EVENT_SUB_PRIORITY, mptcp_sk(subflow->conn), sk, GFP_ATOMIC);
+	mptcp_event(MPTCP_EVENT_SUB_PRIORITY, msk, ssk, GFP_ATOMIC);
 }
 
 void mptcp_pm_mp_fail_received(struct sock *sk, u64 fail_seq)
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index bdba1ddee2a7..cec42bc16d12 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -3128,6 +3128,8 @@ static void mptcp_release_cb(struct sock *sk)
 			__mptcp_set_connected(sk);
 		if (__test_and_clear_bit(MPTCP_ERROR_REPORT, &msk->cb_flags))
 			__mptcp_error_report(sk);
+		if (__test_and_clear_bit(MPTCP_RESET_SCHEDULER, &msk->cb_flags))
+			msk->last_snd = NULL;
 	}
 
 	__mptcp_update_rmem(sk);
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index c8bada4537e2..d59c16d248a4 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -124,6 +124,7 @@
 #define MPTCP_RETRANSMIT	4
 #define MPTCP_FLUSH_JOIN_LIST	5
 #define MPTCP_CONNECTED		6
+#define MPTCP_RESET_SCHEDULER	7
 
 static inline bool before64(__u64 seq1, __u64 seq2)
 {
-- 
2.35.1

[PATCH v2 mptcp-next 3/3] mptcp: reset the packet scheduler on PRIO change

[Paolo Abeni]

Similar to the previous patch, for priority changes
requested by the local PM.

Reported-and-suggested-by: Davide Caratti <dcaratti@redhat.com>
Fixes: 067065422fcd ("mptcp: add the outgoing MP_PRIO support")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
---
 net/mptcp/pm_netlink.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c
index c1f4befb1e45..10a73898c8db 100644
--- a/net/mptcp/pm_netlink.c
+++ b/net/mptcp/pm_netlink.c
@@ -727,6 +727,8 @@ static int mptcp_pm_nl_mp_prio_send_ack(struct mptcp_sock *msk,
 		if (!addresses_equal(&local, addr, addr->port))
 			continue;
 
+		if (subflow->backup != bkup)
+			msk->last_snd = NULL;
 		subflow->backup = bkup;
 		subflow->send_mp_prio = 1;
 		subflow->request_bkup = bkup;
-- 
2.35.1

Re: [PATCH v2 mptcp-next 2/3] mptcp: reset the packet scheduler on incoming MP_PRIO.

[Mat Martineau]

On Tue, 22 Mar 2022, Paolo Abeni wrote:

> When an incoming MP_PRIO option changes the backup
> status of any subflow, we need to reset the packet
> scheduler status, or the next send could keep using
> the previously selected subflow, without taking in account
> the new priorities.
>
> Reported-by: Davide Caratti <dcaratti@redhat.com>
> Fixes: 40453a5c61f4 ("mptcp: add the incoming MP_PRIO support")
> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
> ---
> v1 -> v2:
> - fix backup comparison - Mat

Thanks for the v2, series is ready for the export branch:

Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>

> ---
> net/mptcp/pm.c       | 19 +++++++++++++++----
> net/mptcp/protocol.c |  2 ++
> net/mptcp/protocol.h |  1 +
> 3 files changed, 18 insertions(+), 4 deletions(-)
>
> diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c
> index d0d31d5c198a..c87f8974a1a7 100644
> --- a/net/mptcp/pm.c
> +++ b/net/mptcp/pm.c
> @@ -262,14 +262,25 @@ void mptcp_pm_rm_addr_received(struct mptcp_sock *msk,
> 	spin_unlock_bh(&pm->lock);
> }
>
> -void mptcp_pm_mp_prio_received(struct sock *sk, u8 bkup)
> +void mptcp_pm_mp_prio_received(struct sock *ssk, u8 bkup)
> {
> -	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
> +	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(ssk);
> +	struct sock *sk = subflow->conn;
> +	struct mptcp_sock *msk;
>
> 	pr_debug("subflow->backup=%d, bkup=%d\n", subflow->backup, bkup);
> -	subflow->backup = bkup;
> +	msk = mptcp_sk(sk);
> +	if (subflow->backup != bkup) {
> +		subflow->backup = bkup;
> +		mptcp_data_lock(sk);
> +		if (!sock_owned_by_user(sk))
> +			msk->last_snd = NULL;
> +		else
> +			__set_bit(MPTCP_RESET_SCHEDULER,  &msk->cb_flags);
> +		mptcp_data_unlock(sk);
> +	}
>
> -	mptcp_event(MPTCP_EVENT_SUB_PRIORITY, mptcp_sk(subflow->conn), sk, GFP_ATOMIC);
> +	mptcp_event(MPTCP_EVENT_SUB_PRIORITY, msk, ssk, GFP_ATOMIC);
> }
>
> void mptcp_pm_mp_fail_received(struct sock *sk, u64 fail_seq)
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index bdba1ddee2a7..cec42bc16d12 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -3128,6 +3128,8 @@ static void mptcp_release_cb(struct sock *sk)
> 			__mptcp_set_connected(sk);
> 		if (__test_and_clear_bit(MPTCP_ERROR_REPORT, &msk->cb_flags))
> 			__mptcp_error_report(sk);
> +		if (__test_and_clear_bit(MPTCP_RESET_SCHEDULER, &msk->cb_flags))
> +			msk->last_snd = NULL;
> 	}
>
> 	__mptcp_update_rmem(sk);
> diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
> index c8bada4537e2..d59c16d248a4 100644
> --- a/net/mptcp/protocol.h
> +++ b/net/mptcp/protocol.h
> @@ -124,6 +124,7 @@
> #define MPTCP_RETRANSMIT	4
> #define MPTCP_FLUSH_JOIN_LIST	5
> #define MPTCP_CONNECTED		6
> +#define MPTCP_RESET_SCHEDULER	7
>
> static inline bool before64(__u64 seq1, __u64 seq2)
> {
> -- 
> 2.35.1
>
>
>

--
Mat Martineau
Intel

Re: [PATCH v2 mptcp-next 2/3] mptcp: reset the packet scheduler on incoming MP_PRIO.

[Matthieu Baerts]

Hi Paolo, Mat, Davide,

On 22/03/2022 17:54, Mat Martineau wrote:
> On Tue, 22 Mar 2022, Paolo Abeni wrote:
> 
>> When an incoming MP_PRIO option changes the backup
>> status of any subflow, we need to reset the packet
>> scheduler status, or the next send could keep using
>> the previously selected subflow, without taking in account
>> the new priorities.
>>
>> Reported-by: Davide Caratti <dcaratti@redhat.com>
>> Fixes: 40453a5c61f4 ("mptcp: add the incoming MP_PRIO support")
>> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
>> ---
>> v1 -> v2:
>> - fix backup comparison - Mat
> 
> Thanks for the v2, series is ready for the export branch:
> 
> Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>

Thank you for the patches, reviews, reports and suggestions!

These patches are now in our tree (feat. for net-next) with Mat's RvB tags.

New patches for t/upstream:
- 9d336ad77b77: mptcp: optimize release_cb for the common case
- fadfdaea5a2f: mptcp: reset the packet scheduler on incoming MP_PRIO
- b9854c166093: mptcp: reset the packet scheduler on PRIO change
- Results: 73b94fb27a8c..ff138f44bccd (export)

Builds and tests are now in progress:
https://cirrus-ci.com/github/multipath-tcp/mptcp_net-next/export/20220323T171227
https://github.com/multipath-tcp/mptcp_net-next/actions/workflows/build-validation.yml?query=branch:export



@Davide/Paolo: I didn't merge the new packetdrill tests because there
were still some opened questions. May you have a look if you don't mind? :)

https://github.com/multipath-tcp/packetdrill/pull/82


Cheers,
Matt
-- 
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net