mptcp: improve ADD_ADDR timeout handling
Subsystems (4)
Geliang Tang
reviewerv3 cover mptcp: improve ADD_ADDR timeout handling
From: Geliang Tang <[email protected]>
v3:
- a new patch, remove duplicate sk_reset_timer call
- use __tcp_set_rto() * 2 instead of icsk->icsk_rto
- update selftests.
Based-on: <[email protected]>
v2:
- do not rename add_addr_timeout
- use icsk->icsk_rto instead of tp->srtt_us
- increase the time after each retransmission
- https://patchwork.kernel.org/project/mptcp/patch/40706ad511220729207ccd7cf48e320e4d0d7dea.1754028199.git.tanggeliang@kylinos.cn/
v1:
- https://patchwork.kernel.org/project/mptcp/cover/[email protected]/
Geliang Tang (3):
mptcp: remove duplicate sk_reset_timer call
mptcp: make ADD_ADDR retransmission timeout adaptive
selftests: mptcp: remove add_addr_timeout settings
Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 36 +++++++++++++++---- .../testing/selftests/net/mptcp/mptcp_join.sh | 3 -- 3 files changed, 31 insertions(+), 12 deletions(-)
Signature · 1 line
2.48.1
Geliang Tang
reviewerPATCH v3 1/3 mptcp: remove duplicate sk_reset_timer call
From: Geliang Tang <[email protected]>
sk_reset_timer() was called twice in mptcp_pm_alloc_anno_list. Simplify the code by using a 'goto' statement to eliminate the duplication.
--- net/mptcp/pm.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)
net/mptcp/pm.c +3 -5
@@ -348,7 +348,6 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, {struct mptcp_pm_add_entry *add_entry =NULL ;struct sock *sk = (struct sock *)msk; -struct net *net =sock_net (sk);lockdep_assert_held (&msk->pm.lock);
@@ -358,9 +357,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,if (WARN_ON_ONCE (mptcp_pm_is_kernel (msk)))return false; -sk_reset_timer (sk, &add_entry->add_timer, - jiffies +mptcp_get_add_addr_timeout (net)); -return true; + goto out; } add_entry =kmalloc (sizeof (*add_entry),GFP_ATOMIC );
@@ -374,8 +371,9 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, add_entry->retrans_times =0 ;timer_setup (&add_entry->add_timer, mptcp_pm_add_timer,0 ); +out:sk_reset_timer (sk, &add_entry->add_timer, - jiffies +mptcp_get_add_addr_timeout (net)); + jiffies +mptcp_get_add_addr_timeout (sock_net (sk)));return true; }
Signature · 1 line
2.48.1
Matthieu Baerts
maintainerRe: [PATCH mptcp-next v3 1/3] mptcp: remove duplicate sk_reset_timer call
Hi Geliang,
Quoted 23 lines · Geliang Tang
From: Geliang Tang <[email protected]>
sk_reset_timer()was called twice inmptcp_pm_alloc_anno_list. Simplify the code by using a 'goto' statement to eliminate the duplication.Signed-off-by Geliang Tang <[email protected]>--- net/mptcp/pm.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-)net/mptcp/pm.c +0
-1@@ -348,7 +348,6 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, {struct mptcp_pm_add_entry *add_entry =NULL ;struct sock *sk = (struct sock *)msk; -struct net *net =sock_net (sk);lockdep_assert_held (&msk->pm.lock);
net/mptcp/pm.c bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, · line 360
Quoted 7 lines
net/mptcp/pm.c +1
-3@@ -358,9 +357,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,if (WARN_ON_ONCE (mptcp_pm_is_kernel (msk)))return false; -sk_reset_timer (sk, &add_entry->add_timer, - jiffies +mptcp_get_add_addr_timeout (net)); -return true; + goto out;
Good idea, but maybe better to have a more explicit label, e.g. reset_timer.
Quoted 4 lines
net/mptcp/pm.c +0
-0} add_entry =kmalloc (sizeof (*add_entry),GFP_ATOMIC );
net/mptcp/pm.c bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, · line 379
Quoted 10 lines
net/mptcp/pm.c +2
-1@@ -374,8 +371,9 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, add_entry->retrans_times =0 ;timer_setup (&add_entry->add_timer, mptcp_pm_add_timer,0 ); +out:sk_reset_timer (sk, &add_entry->add_timer, - jiffies +mptcp_get_add_addr_timeout (net)); + jiffies +mptcp_get_add_addr_timeout (sock_net (sk)));return true; }
Cheers,
Matt
Signature · 1 line
Sponsored by the NGI0 Core fund.
Geliang Tang
reviewerPATCH v3 2/3 mptcp: make ADD_ADDR retransmission timeout adaptive
In reply to Geliang TangFrom: Geliang Tang <[email protected]>
Replace the fixed ADD_ADDR retransmission timeout with an adaptive mechanism that uses the maximum RTO among existing subflows. This improves responsiveness when establishing new subflows while maintaining an upper bound through the configured add_addr_timeout value.
Key changes:
1. Compute timeout as min(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows exist
4. Update documentation to clarify add_addr_timeout is now the maximum value
--- Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30 ++++++++++++++++++++--- 2 files changed, 29 insertions(+), 5 deletions(-)
Documentation/networking/mptcp-sysctl.rst +2 -2
@@ -8,8 +8,8 @@ MPTCP Sysfs variables =============================== add_addr_timeout - INTEGER (seconds) - Set the timeout after which an ADD_ADDR control message will be - resent to an MPTCP peer that has not acknowledged a previous + Set the maximum value of timeout after which an ADD_ADDR control message + will be resent to an MPTCP peer that has not acknowledged a previous ADD_ADDR message. Do not retransmit if set to 0.
net/mptcp/pm.c +27 -3
@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk,return -EINVAL ; } +static unsigned int mptcp_adjust_add_addr_timeout (struct mptcp_sock *msk) +{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ; +unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { +struct sock *ssk =mptcp_subflow_tcp_sock (subflow); +unsigned int subflow_rto; + + subflow_rto =__tcp_set_rto (tcp_sk (ssk)) *2 ; +if (subflow_rto > max_rto) + max_rto = subflow_rto; + } + +if (max_rto && max_rto < timeout) + timeout = max_rto; + +return timeout; +} +static void mptcp_pm_add_timer (struct timer_list *timer) { struct mptcp_pm_add_entry *entry = timer_container_of(entry, timer,
@@ -292,7 +316,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer) goto out; } - timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + timeout =mptcp_adjust_add_addr_timeout (msk);if (!timeout) goto out;
@@ -307,7 +331,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer)if (entry->retrans_times <ADD_ADDR_RETRANS_MAX )sk_reset_timer (sk, timer, - jiffies + timeout); + jiffies + (timeout << entry->retrans_times));spin_unlock_bh (&msk->pm.lock);
@@ -373,7 +397,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,timer_setup (&add_entry->add_timer, mptcp_pm_add_timer,0 ); out:sk_reset_timer (sk, &add_entry->add_timer, - jiffies +mptcp_get_add_addr_timeout (sock_net (sk))); + jiffies +mptcp_adjust_add_addr_timeout (msk));return true; }
Signature · 1 line
2.48.1
Matthieu Baerts
maintainerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Hi Geliang,
Quoted 49 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an adaptive mechanism that uses the maximum RTO among existing subflows. This improves responsiveness when establishing new subflows while maintaining an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows exist
4. Update documentation to clarifyadd_addr_timeoutis now the maximum valueReviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>--- Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30 ++++++++++++++++++++--- 2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +2
-2@@ -8,8 +8,8 @@ MPTCP Sysfs variables =============================== add_addr_timeout - INTEGER (seconds) - Set the timeout after which an ADD_ADDR control message will be - resent to an MPTCP peer that has not acknowledged a previous + Set the maximum value of timeout after which an ADD_ADDR control message + will be resent to an MPTCP peer that has not acknowledged a previous ADD_ADDR message. Do not retransmit if set to 0.net/mptcp/pm.c +4
-0@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk,return -EINVAL ; } +static unsigned int mptcp_adjust_add_addr_timeout (struct mptcp_sock *msk) +{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk;
net/mptcp/pm.c int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk, · line 284
Quoted 10 lines
net/mptcp/pm.c +10
-0+unsigned int max_rto =0 ; +unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { +struct sock *ssk =mptcp_subflow_tcp_sock (subflow); +unsigned int subflow_rto; + + subflow_rto =__tcp_set_rto (tcp_sk (ssk)) *2 ;
Why "* 2"? Is it what TCP is doing?
(Probably best to add a comment here above, and in the commit message.)
(Also, I guess the compiler will do the optimisation, but you could
write: ">> 1" instead of "* 2".
Quoted 32 lines
net/mptcp/pm.c +12
-2+if (subflow_rto > max_rto) + max_rto = subflow_rto; + } + +if (max_rto && max_rto < timeout) + timeout = max_rto; + +return timeout; +} +static void mptcp_pm_add_timer (struct timer_list *timer) { struct mptcp_pm_add_entry *entry = timer_container_of(entry, timer,@@ -292,7 +316,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer) goto out; } - timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + timeout =mptcp_adjust_add_addr_timeout (msk);if (!timeout) goto out;@@ -307,7 +331,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer)if (entry->retrans_times <ADD_ADDR_RETRANS_MAX )sk_reset_timer (sk, timer, - jiffies + timeout); + jiffies + (timeout << entry->retrans_times));spin_unlock_bh (&msk->pm.lock);
net/mptcp/pm.c bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, · line 403
Quoted 8 lines
net/mptcp/pm.c +1
-1@@ -373,7 +397,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,timer_setup (&add_entry->add_timer, mptcp_pm_add_timer,0 ); out:sk_reset_timer (sk, &add_entry->add_timer, - jiffies +mptcp_get_add_addr_timeout (sock_net (sk))); + jiffies +mptcp_adjust_add_addr_timeout (msk));return true; }
Cheers,
Matt
Signature · 1 line
Sponsored by the NGI0 Core fund.
Geliang Tang
reviewerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Hi Matt,
Quoted 77 lines · Matthieu Baerts
Hi Geliang,
Quoted 68 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an adaptive mechanism that uses the maximum RTO among existing subflows. This improves responsiveness when establishing new subflows while maintaining
an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows exist
4. Update documentation to clarifyadd_addr_timeoutis now the maximum
valueReviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>---
Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30
++++++++++++++++++++---2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +0
-0@@ -8,8 +8,8 @@ MPTCP Sysfs variables===============================
add_addr_timeout - INTEGER (seconds)Documentation/networking/mptcp-sysctl.rst +0
-1- Set the timeout after which an ADD_ADDR control messagewill be
Documentation/networking/mptcp-sysctl.rst +0
-1- resent to an MPTCP peer that has not acknowledged aprevious
Documentation/networking/mptcp-sysctl.rst +1
-0+ Set the maximum value of timeout after which an ADD_ADDRcontrol message
Documentation/networking/mptcp-sysctl.rst +1
-0+ will be resent to an MPTCP peer that has not acknowledgeda previous
ADD_ADDRmessage.
Do not retransmit if set to 0.net/mptcp/pm.c +0
-0@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(structmptcp_sock *msk, return -EINVAL; }
+static unsigned intmptcp_adjust_add_addr_timeout(struct mptcp_sock *msk)net/mptcp/pm.c +10
-0+{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ; +unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { + struct sock *ssk =
mptcp_subflow_tcp_sock(subflow);net/mptcp/pm.c int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk, · line 284
Quoted 3 lines
net/mptcp/pm.c +3
-0+unsigned int subflow_rto; + + subflow_rto =__tcp_set_rto (tcp_sk (ssk)) *2 ;Why "* 2"? Is it what TCP is doing?
Quoted 5 lines
(Probably best to add a comment here above, and in the commit message.)
(Also, I guess the compiler will do the optimisation, but you could
write: ">> 1" instead of "* 2".
Yes, "<< 1" is much better. I added a comment here:
+ /*
+ * icsk->icsk_rto is not used directly since + * it's not updated frequently enough. + * + * subflow_rto is twice of rto, and allow + * some tolerance for add_addr echo. + */
+ subflow_rto = __tcp_set_rto(tcp_sk(ssk)) << 1;
Is this okay?
Thanks,
-Geliang
Quoted 51 lines
Quoted 44 lines
net/mptcp/pm.c +10
-0+if (subflow_rto > max_rto) + max_rto = subflow_rto; + } + +if (max_rto && max_rto < timeout) + timeout = max_rto; + +return timeout; +} +static void
mptcp_pm_add_timer(struct timer_list *timer){
structmptcp_pm_add_entry*entry = timer_container_of(entry, timer,net/mptcp/pm.c +0
-0@@ -292,7 +316,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer) goto out; }net/mptcp/pm.c +1
-1- timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + timeout =mptcp_adjust_add_addr_timeout (msk);if (!timeout)
goto out;net/mptcp/pm.c +0
-0@@ -307,7 +331,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer)
if (entry->retrans_times<ADD_ADDR_RETRANS_MAX) sk_reset_timer(sk, timer,net/mptcp/pm.c +1
-1- jiffies + timeout); + jiffies + (timeout << entry-Quoted 1 line
retrans_times));
spin_unlock_bh(&msk->pm.lock);net/mptcp/pm.c +0
-0@@ -373,7 +397,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk, timer_setup(&add_entry->add_timer, mptcp_pm_add_timer, 0);out:
sk_reset_timer(sk,&add_entry->add_timer,net/mptcp/pm.c +0
-1- jiffies +mptcp_get_add_addr_timeout(sock_net(sk)));
net/mptcp/pm.c +1
-0+ jiffies +Quoted 4 lines
mptcp_adjust_add_addr_timeout(msk));
return true;
}Cheers,
Matt
Matthieu Baerts
maintainerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Hi Geliang,
Quoted 97 lines · Geliang Tang
Hi Matt,
Quoted 77 lines · Matthieu Baerts
Hi Geliang,
Quoted 68 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an adaptive mechanism that uses the maximum RTO among existing subflows. This improves responsiveness when establishing new subflows while maintaining
an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows exist
4. Update documentation to clarifyadd_addr_timeoutis now the maximum
valueReviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>---
Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30
++++++++++++++++++++---2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +0
-0@@ -8,8 +8,8 @@ MPTCP Sysfs variables===============================
add_addr_timeout - INTEGER (seconds)Documentation/networking/mptcp-sysctl.rst +0
-1- Set the timeout after which an ADD_ADDR control messagewill be
Documentation/networking/mptcp-sysctl.rst +0
-1- resent to an MPTCP peer that has not acknowledged aprevious
Documentation/networking/mptcp-sysctl.rst +1
-0+ Set the maximum value of timeout after which an ADD_ADDRcontrol message
Documentation/networking/mptcp-sysctl.rst +1
-0+ will be resent to an MPTCP peer that has not acknowledgeda previous
ADD_ADDRmessage.
Do not retransmit if set to 0.net/mptcp/pm.c +0
-0@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(structmptcp_sock *msk, return -EINVAL; }
+static unsigned intmptcp_adjust_add_addr_timeout(struct mptcp_sock *msk)net/mptcp/pm.c +10
-0+{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ; +unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { + struct sock *ssk =
mptcp_subflow_tcp_sock(subflow);net/mptcp/pm.c int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk, · line 284
Quoted 3 lines
net/mptcp/pm.c +3
-0+unsigned int subflow_rto; + + subflow_rto =__tcp_set_rto (tcp_sk (ssk)) *2 ;Why "* 2"? Is it what TCP is doing?
Quoted 5 lines
(Probably best to add a comment here above, and in the commit message.)
(Also, I guess the compiler will do the optimisation, but you could
write: ">> 1" instead of "* 2".Yes, "<< 1" is much better. I added a comment here:
(quoted diff) +8
-0+ /* + * icsk->icsk_rto is not used directly since + * it's not updated frequently enough. + * + * subflow_rto is twice of rto, and allow + * some tolerance for add_addr echo. + */ + subflow_rto = __tcp_set_rto(tcp_sk(ssk)) << 1;
Quoted 1 line
Is this okay?
No :)
I don't think the comment explain why it is twice the default TCP RTO. Why having an exception here for the ADD_ADDR? I mean: I'm OK with the idea of doubling the RTO, but only if there is a reason behind that :)
Cheers,
Matt
Signature · 1 line
Sponsored by the NGI0 Core fund.
Geliang Tang
reviewerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Hi Matt,
Quoted 110 lines · Matthieu Baerts
Hi Geliang,
Quoted 105 lines · Geliang Tang
Hi Matt,
Quoted 84 lines · Matthieu Baerts
Hi Geliang,
Quoted 75 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an adaptive
mechanism that uses the maximum RTO among existing subflows. This
improves responsiveness when establishing new subflows while maintaining
an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows exist
4. Update documentation to clarifyadd_addr_timeoutis now the maximum
valueCloses:
https://github.com/multipath-tcp/mptcp_net-next/issues/576Reviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>---
Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30
++++++++++++++++++++---2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +0
-0@@ -8,8 +8,8 @@ MPTCP Sysfs variables===============================
add_addr_timeout - INTEGER (seconds)Documentation/networking/mptcp-sysctl.rst +0
-1- Set the timeout after which an ADD_ADDR controlmessage
will beDocumentation/networking/mptcp-sysctl.rst +0
-1- resent to an MPTCP peer that has not acknowledged aprevious
Documentation/networking/mptcp-sysctl.rst +1
-0+ Set the maximum value of timeout after which an
ADD_ADDR
control messageDocumentation/networking/mptcp-sysctl.rst +1
-0+ will be resent to an MPTCP peer that has notacknowledged
a previous
ADD_ADDRmessage.
Do not retransmit if set to 0.net/mptcp/pm.c +0
-0@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(structmptcp_sock *msk, return -EINVAL; }
+static unsigned intmptcp_adjust_add_addr_timeout(struct mptcp_sock *msk)net/mptcp/pm.c +10
-0+{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ; +unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { + struct sock *ssk =
mptcp_subflow_tcp_sock(subflow);net/mptcp/pm.c int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk, · line 284
Quoted 3 lines
net/mptcp/pm.c +3
-0+unsigned int subflow_rto; + + subflow_rto =__tcp_set_rto (tcp_sk (ssk)) *2 ;Why "* 2"? Is it what TCP is doing?
Quoted 6 lines
(Probably best to add a comment here above, and in the commit message.)
(Also, I guess the compiler will do the optimisation, but you could
write: ">> 1" instead of "* 2".Yes, "<< 1" is much better. I added a comment here:
+ /*
+ *icsk->icsk_rtois not used directly since + * it's not updated frequently enough. + *
+ *subflow_rtois twice of rto, and allow + * some tolerance foradd_addrecho. + */
+subflow_rto= __tcp_set_rto(tcp_sk(ssk)) << 1;Quoted 1 line
Is this okay?
Quoted 8 lines
No :)
I don't think the comment explain why it is twice the default TCP RTO.
Why having an exception here for theADD_ADDR? I mean: I'm OK with the
idea of doubling the RTO, but only if there is a reason behind that :)
How about this one (answered by DeepSeek, an AI from China):
/*
* icsk->icsk_rto is not used directly since it's not
* updated frequently enough.
*
* Setting subflow_rto to twice of RTO provides a safety
* buffer for RTT estimation variance, kernel RTO update
* delays, and a full transmission retry cycle while
* minimizing false failures.
*/
subflow_rto = __tcp_set_rto(tcp_sk(ssk)) << 1;
Thanks,
-Geliang
Quoted 3 lines
Cheers,
Matt
Matthieu Baerts
maintainerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Hi Geliang,
Quoted 122 lines · Geliang Tang
Hi Matt,
Quoted 110 lines · Matthieu Baerts
Hi Geliang,
Quoted 105 lines · Geliang Tang
Hi Matt,
Quoted 84 lines · Matthieu Baerts
Hi Geliang,
Quoted 75 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an adaptive
mechanism that uses the maximum RTO among existing subflows. This
improves responsiveness when establishing new subflows while maintaining
an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows exist
4. Update documentation to clarifyadd_addr_timeoutis now the maximum
valueCloses:
https://github.com/multipath-tcp/mptcp_net-next/issues/576Reviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>---
Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30
++++++++++++++++++++---2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +0
-0@@ -8,8 +8,8 @@ MPTCP Sysfs variables===============================
add_addr_timeout - INTEGER (seconds)Documentation/networking/mptcp-sysctl.rst +0
-1- Set the timeout after which an ADD_ADDR controlmessage
will beDocumentation/networking/mptcp-sysctl.rst +0
-1- resent to an MPTCP peer that has not acknowledged aprevious
Documentation/networking/mptcp-sysctl.rst +1
-0+ Set the maximum value of timeout after which an
ADD_ADDR
control messageDocumentation/networking/mptcp-sysctl.rst +1
-0+ will be resent to an MPTCP peer that has notacknowledged
a previous
ADD_ADDRmessage.
Do not retransmit if set to 0.net/mptcp/pm.c +0
-0@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(structmptcp_sock *msk, return -EINVAL; }
+static unsigned intmptcp_adjust_add_addr_timeout(struct mptcp_sock *msk)net/mptcp/pm.c +10
-0+{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ; +unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { + struct sock *ssk =
mptcp_subflow_tcp_sock(subflow);net/mptcp/pm.c int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk, · line 284
Quoted 3 lines
net/mptcp/pm.c +3
-0+unsigned int subflow_rto; + + subflow_rto =__tcp_set_rto (tcp_sk (ssk)) *2 ;Why "* 2"? Is it what TCP is doing?
Quoted 6 lines
(Probably best to add a comment here above, and in the commit message.)
(Also, I guess the compiler will do the optimisation, but you could
write: ">> 1" instead of "* 2".Yes, "<< 1" is much better. I added a comment here:
+ /*
+ *icsk->icsk_rtois not used directly since + * it's not updated frequently enough. + *
+ *subflow_rtois twice of rto, and allow + * some tolerance foradd_addrecho. + */
+subflow_rto= __tcp_set_rto(tcp_sk(ssk)) << 1;Quoted 1 line
Is this okay?
Quoted 8 lines
No :)
I don't think the comment explain why it is twice the default TCP RTO.
Why having an exception here for theADD_ADDR? I mean: I'm OK with the
idea of doubling the RTO, but only if there is a reason behind that :)
Quoted 5 lines
How about this one (answered by DeepSeek, an AI from China):
/*
*icsk->icsk_rtois not used directly since it's not
* updated frequently enough.
What about: (...) since it might not have been updated recently.
Quoted 4 lines
* Setting
subflow_rtoto twice of RTO provides a safety
* buffer for RTT estimation variance, kernel RTO update
* delays, and a full transmission retry cycle while
* minimizing false failures.
A bit long and not clear (typical AI, making commit messages too verbose, it almost force me to use one to give me the summary :) )
What about:
Setting this RTO to twice of TCP one, not to be too agressive for this notification that might take longer to process than "simple" data.
Still, is icsk_rto not already "long" enough by default? I mean: when I see your description in the related packetdrill PR [1] where the delay is supposed to be 0.01 sec, but "subflow_rto" (or was it isck_rto?) is 0.65 sec, that's quite different, and half of it would be more than enough :)
But maybe that's different when it happens later during the connection?
[1] https://github.com/multipath-tcp/packetdrill/pull/166
Cheers,
Matt
Signature · 1 line
Sponsored by the NGI0 Core fund.
Re: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Quoted 147 lines · Matthieu Baerts
Hi Geliang,
Quoted 122 lines · Geliang Tang
Hi Matt,
Quoted 110 lines · Matthieu Baerts
Hi Geliang,
Quoted 105 lines · Geliang Tang
Hi Matt,
Quoted 84 lines · Matthieu Baerts
Hi Geliang,
Quoted 68 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an adaptive
mechanism that uses the maximum RTO among existing subflows. This
improves responsiveness when establishing new subflows while maintaining
an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows exist
4. Update documentation to clarifyadd_addr_timeoutis now the maximum
valueCloses:
https://github.com/multipath-tcp/mptcp_net-next/issues/576Reviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>--- Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30 ++++++++++++++++++++--- 2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +0
-1@@ -8,8 +8,8 @@ MPTCP Sysfs variables =============================== add_addr_timeout - INTEGER (seconds) - Set the timeout after which an ADD_ADDR controlmessage
will be
- resent to an MPTCP peer that has not acknowledged a previous
+ Set the maximum value of timeout after which anADD_ADDR
control message
+ will be resent to an MPTCP peer that has not acknowledged
a previous
ADD_ADDRmessage.Do not retransmit if set to 0.
net/mptcp/pm.c +5
-0@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk, return -EINVAL ; } +static unsigned int mptcp_adjust_add_addr_timeout (struct mptcp_sock *msk) +{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ;net/mptcp/pm.c int mptcp_pm_mp_prio_send_ack(struct · line 287
Quoted 10 lines
net/mptcp/pm.c +9
-0+unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { +struct sock *ssk =mptcp_subflow_tcp_sock (subflow); +unsigned int subflow_rto; + + subflow_rto =__tcp_set_rto (tcp_sk (ssk)) *2 ;Why "* 2"? Is it what TCP is doing?
Quoted 6 lines
(Probably best to add a comment here above, and in the commit message.)
(Also, I guess the compiler will do the optimisation, but you could
write: ">> 1" instead of "* 2".Yes, "<< 1" is much better. I added a comment here:
(quoted diff) +8
-0+ /* + * icsk->icsk_rto is not used directly since + * it's not updated frequently enough. + * + * subflow_rto is twice of rto, and allow + * some tolerance for add_addr echo. + */ + subflow_rto = __tcp_set_rto(tcp_sk(ssk)) << 1;Quoted 1 line
Is this okay?
Quoted 8 lines
No :)
I don't think the comment explain why it is twice the default TCP RTO.
Why having an exception here for theADD_ADDR? I mean: I'm OK with the
idea of doubling the RTO, but only if there is a reason behind that :)Quoted 5 lines
How about this one (answered by DeepSeek, an AI from China):
/*
*icsk->icsk_rtois not used directly since it's not
* updated frequently enough.What about: (...) since it might not have been updated recently.
Quoted 4 lines
* Setting
subflow_rtoto twice of RTO provides a safety
* buffer for RTT estimation variance, kernel RTO update
* delays, and a full transmission retry cycle while
* minimizing false failures.A bit long and not clear (typical AI, making commit messages too verbose, it almost force me to use one to give me the summary :) )
What about:
Setting this RTO to twice of TCP one, not to be too agressive for this notification that might take longer to process than "simple" data.
Quoted 7 lines
Still, is
icsk_rtonot already "long" enough by default? I mean: when I see your description in the related packetdrill PR [1] where the delay is supposed to be 0.01 sec, but "subflow_rto" (or was itisck_rto?) is 0.65 sec, that's quite different, and half of it would be more than enough :)But maybe that's different when it happens later during the connection?
Why wouldn't icsk_rto not be good enough ? We just need a reasonable number here. And, I must say that 2 x __tcp_set_rto() seems a bit arbitrary. Let's just use icsk_rto. Does it really matter that it is not updated very frequently ?
Christoph
Quoted 8 lines
[1] https://github.com/multipath-tcp/packetdrill/pull/166
Cheers,
Matt
--
Sponsored by the NGI0 Core fund.
Geliang Tang
reviewerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Hi Christoph, Matt,
Quoted 172 lines · Christoph Paasch
Quoted 170 lines · Matthieu Baerts
Hi Geliang,
Quoted 134 lines · Geliang Tang
Hi Matt,
Quoted 121 lines · Matthieu Baerts
Hi Geliang,
Quoted 114 lines · Geliang Tang
Hi Matt,
Quoted 90 lines · Matthieu Baerts
Hi Geliang,
Quoted 74 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an adaptive
mechanism that uses the maximum RTO among existing subflows.
This
improves responsiveness when establishing new subflows while
maintaining
an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows
exist
4. Update documentation to clarifyadd_addr_timeoutis now the
maximum
valueCloses:
https://github.com/multipath-tcp/mptcp_net-next/issues/576Reviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>---
Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30
++++++++++++++++++++---2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +0
-0@@ -8,8 +8,8 @@ MPTCP Sysfs variables===============================
add_addr_timeout - INTEGER (seconds)
- Set the timeout after which anADD_ADDRcontrol message
will be
- resent to an MPTCP peer that has not acknowledged a previous
+ Set the maximum value of timeout after which anADD_ADDR
control message
+ will be resent to an MPTCP peer that has not acknowledged
a previous
ADD_ADDRmessage.Do not retransmit if set to 0.
net/mptcp/pm.c +5
-0@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(structmptcp_sock *msk, return -EINVAL; } +static unsigned int mptcp_adjust_add_addr_timeout (struct mptcp_sock *msk) +{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ;net/mptcp/pm.c int mptcp_pm_mp_prio_send_ack(struct · line 287
Quoted 10 lines
net/mptcp/pm.c +9
-0+unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { + struct sock *ssk =mptcp_subflow_tcp_sock (subflow); + unsigned int subflow_rto; + + subflow_rto = __tcp_set_rto(tcp_sk(ssk)) * 2;Why "* 2"? Is it what TCP is doing?
Quoted 8 lines
(Probably best to add a comment here above, and in the commit
message.)(Also, I guess the compiler will do the optimisation, but you
could
write: ">> 1" instead of "* 2".Yes, "<< 1" is much better. I added a comment here:
+ /*
+ *icsk->icsk_rtois not used directly since + * it's not updated frequently enough. + *
+ *subflow_rtois twice of rto, and allow + * some tolerance foradd_addrecho. + */
+subflow_rto= __tcp_set_rto(tcp_sk(ssk)) << 1;Quoted 1 line
Is this okay?
No :)
Quoted 9 lines
I don't think the comment explain why it is twice the default TCP
RTO.
Why having an exception here for theADD_ADDR? I mean: I'm OK with
the
idea of doubling the RTO, but only if there is a reason behind that
:)Quoted 5 lines
How about this one (answered by DeepSeek, an AI from China):
/*
*icsk->icsk_rtois not used directly since it's not
* updated frequently enough.What about: (...) since it might not have been updated recently.
Quoted 4 lines
* Setting
subflow_rtoto twice of RTO provides a safety
* buffer for RTT estimation variance, kernel RTO update
* delays, and a full transmission retry cycle while
* minimizing false failures.A bit long and not clear (typical AI, making commit messages too verbose, it almost force me to use one to give me the summary :) )
What about:
Setting this RTO to twice of TCP one, not to be too agressive for this
notification that might take longer to process than "simple" data.Still, is
icsk_rtonot already "long" enough by default? I mean:
when I
see your description in the related packetdrill PR [1] where the delay
is supposed to be 0.01 sec, but "subflow_rto" (or was itisck_rto?) is
0.65 sec, that's quite different, and half of it would be more than enough :)
Quoted 7 lines
Quoted 2 lines
But maybe that's different when it happens later during the connection?
Why wouldn't
icsk_rtonot be good enough ? We just need a reasonable number here. And, I must say that 2 x__tcp_set_rto()seems a bit arbitrary. Let's just useicsk_rto. Does it really matter that it is not updated very frequently ?
Sure, I'll update this in v4, just using icsk->icsk_rto.
Thanks,
-Geliang
Quoted 12 lines
Christoph
Quoted 8 lines
[1] https://github.com/multipath-tcp/packetdrill/pull/166
Cheers,
Matt
--
Sponsored by the NGI0 Core fund.
Matthieu Baerts
maintainerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
In reply to Christoph PaaschHi Christoph,
Quoted 154 lines · Christoph Paasch
Quoted 153 lines · Matthieu Baerts
Hi Geliang,
Quoted 122 lines · Geliang Tang
Hi Matt,
Quoted 110 lines · Matthieu Baerts
Hi Geliang,
Quoted 105 lines · Geliang Tang
Hi Matt,
Quoted 84 lines · Matthieu Baerts
Hi Geliang,
Quoted 68 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an adaptive
mechanism that uses the maximum RTO among existing subflows. This
improves responsiveness when establishing new subflows while maintaining
an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows exist
4. Update documentation to clarifyadd_addr_timeoutis now the maximum
valueCloses:
https://github.com/multipath-tcp/mptcp_net-next/issues/576Reviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>--- Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30 ++++++++++++++++++++--- 2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +0
-1@@ -8,8 +8,8 @@ MPTCP Sysfs variables =============================== add_addr_timeout - INTEGER (seconds) - Set the timeout after which an ADD_ADDR controlmessage
will be
- resent to an MPTCP peer that has not acknowledged a previous
+ Set the maximum value of timeout after which anADD_ADDR
control message
+ will be resent to an MPTCP peer that has not acknowledged
a previous
ADD_ADDRmessage.Do not retransmit if set to 0.
net/mptcp/pm.c +5
-0@@ -268,6 +268,30 @@ int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk, return -EINVAL ; } +static unsigned int mptcp_adjust_add_addr_timeout (struct mptcp_sock *msk) +{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ;net/mptcp/pm.c int mptcp_pm_mp_prio_send_ack(struct · line 287
Quoted 10 lines
net/mptcp/pm.c +9
-0+unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { +struct sock *ssk =mptcp_subflow_tcp_sock (subflow); +unsigned int subflow_rto; + + subflow_rto =__tcp_set_rto (tcp_sk (ssk)) *2 ;Why "* 2"? Is it what TCP is doing?
Quoted 6 lines
(Probably best to add a comment here above, and in the commit message.)
(Also, I guess the compiler will do the optimisation, but you could
write: ">> 1" instead of "* 2".Yes, "<< 1" is much better. I added a comment here:
(quoted diff) +8
-0+ /* + * icsk->icsk_rto is not used directly since + * it's not updated frequently enough. + * + * subflow_rto is twice of rto, and allow + * some tolerance for add_addr echo. + */ + subflow_rto = __tcp_set_rto(tcp_sk(ssk)) << 1;Quoted 1 line
Is this okay?
Quoted 8 lines
No :)
I don't think the comment explain why it is twice the default TCP RTO.
Why having an exception here for theADD_ADDR? I mean: I'm OK with the
idea of doubling the RTO, but only if there is a reason behind that :)Quoted 5 lines
How about this one (answered by DeepSeek, an AI from China):
/*
*icsk->icsk_rtois not used directly since it's not
* updated frequently enough.What about: (...) since it might not have been updated recently.
Quoted 4 lines
* Setting
subflow_rtoto twice of RTO provides a safety
* buffer for RTT estimation variance, kernel RTO update
* delays, and a full transmission retry cycle while
* minimizing false failures.A bit long and not clear (typical AI, making commit messages too verbose, it almost force me to use one to give me the summary :) )
What about:
Setting this RTO to twice of TCP one, not to be too agressive for this notification that might take longer to process than "simple" data.
Still, is
icsk_rtonot already "long" enough by default? I mean: when I see your description in the related packetdrill PR [1] where the delay is supposed to be 0.01 sec, but "subflow_rto" (or was itisck_rto?) is 0.65 sec, that's quite different, and half of it would be more than enough :)
Quoted 4 lines
Quoted 1 line
But maybe that's different when it happens later during the connection?
Why wouldn't
icsk_rtonot be good enough ? We just need a reasonable number here.
I didn't have time to investigate in between, but is icsk_rto not updated only when needed (@Geliang: did you check?). So could it be possible that at the beginning of the connection, the value is not correct?
If yes, why not using __tcp_set_rto()?
Quoted 3 lines
And, I must say that 2 x
__tcp_set_rto()seems a bit arbitrary. Let's just useicsk_rto. Does it really matter that it is not updated very frequently ?
Maybe yes.
@Geliang: but before sending a new version, please check my comment:
https://github.com/multipath-tcp/packetdrill/pull/166
Especially this part:
Quoted 9 lines
0.65s looks a lot to me: The 3rd ACK at line 15 should be injected with a delay of 0.01s.
What are the values of
icsk->icsk_rto,tp->srtt_us,tp->rttvar_us,__tcp_set_rto(tp)and the one you set for the timer when scheduling each retransmissions and when running this test?(for the rto value in jiffies, please convert them to usec like the others →
jiffies_to_usecs())
Sharing such info when running the packetdrill tests using a non debug kernel (and ideally also sharing the output of packetdrill, when running it with the extra -vvvv), would help better understand the situation here.
Cheers,
Matt
Signature · 1 line
Sponsored by the NGI0 Core fund.
Geliang Tang
reviewerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Hi Matt, Christoph,
Quoted 220 lines · Matthieu Baerts
Hi Christoph,
Quoted 180 lines · Christoph Paasch
Quoted 178 lines · Matthieu Baerts
Hi Geliang,
Quoted 138 lines · Geliang Tang
Hi Matt,
Quoted 125 lines · Matthieu Baerts
Hi Geliang,
Quoted 118 lines · Geliang Tang
Hi Matt,
Quoted 93 lines · Matthieu Baerts
Hi Geliang,
Quoted 83 lines · Geliang Tang
From: Geliang Tang <[email protected]>
Replace the fixed
ADD_ADDRretransmission timeout with an
adaptive
mechanism that uses the maximum RTO among existing subflows.
This
improves responsiveness when establishing new subflows while
maintaining
an upper bound through the configuredadd_addr_timeoutvalue.Key changes:
1. Compute timeout asmin(max_subflow_RTO, configured_max_timeout)
2. Apply exponential backoff based on retransmission count
3. Maintain fallback to configured max timeout when no subflows
exist
4. Update documentation to clarifyadd_addr_timeoutis now the
maximum
valueCloses:
https://github.com/multipath-tcp/mptcp_net-next/issues/576Reviewed-by Christoph Paasch <[email protected]>Signed-off-by Geliang Tang <[email protected]>---
Documentation/networking/mptcp-sysctl.rst | 4 +-- net/mptcp/pm.c | 30
++++++++++++++++++++---2 files changed, 29 insertions(+), 5 deletions(-)Documentation/networking/mptcp-sysctl.rst +0
-0@@ -8,8 +8,8 @@ MPTCP Sysfs variables===============================
add_addr_timeout - INTEGER (seconds)
- Set the timeout after which anADD_ADDRcontrol message
will be
- resent to an MPTCP peer that has not acknowledged a previous
+ Set the maximum value of timeout after which anADD_ADDR
control message
+ will be resent to an MPTCP peer that has not acknowledged
a previous
ADD_ADDRmessage.Do not retransmit if set to 0.
net/mptcp/pm.c +0
-0@@ -268,6 +268,30 @@ intmptcp_pm_mp_prio_send_ack(struct
mptcp_sock*msk,
return-EINVAL;
}+static unsigned int
mptcp_adjust_add_addr_timeout(structmptcp_sock*msk)net/mptcp/pm.c +10
-0+{ +struct mptcp_subflow_context *subflow; +struct sock *sk = (struct sock *)msk; +unsigned int max_rto =0 ; +unsigned int timeout; + + timeout =mptcp_get_add_addr_timeout (sock_net (sk)); + +mptcp_for_each_subflow (msk, subflow) { + struct sock *ssk =Quoted 4 lines
mptcp_subflow_tcp_sock(subflow);
+ unsigned intsubflow_rto;
+
+subflow_rto= __tcp_set_rto(tcp_sk(ssk)) * 2;Why "* 2"? Is it what TCP is doing?
Quoted 8 lines
(Probably best to add a comment here above, and in the commit
message.)(Also, I guess the compiler will do the optimisation, but you
could
write: ">> 1" instead of "* 2".Yes, "<< 1" is much better. I added a comment here:
+ /*
+ *icsk->icsk_rtois not used directly since
+ * it's not updated frequently enough. + *
+ *subflow_rtois twice of rto, and allow + * some tolerance foradd_addrecho. + */
+subflow_rto= __tcp_set_rto(tcp_sk(ssk)) << 1;Quoted 1 line
Is this okay?
No :)
Quoted 9 lines
I don't think the comment explain why it is twice the default TCP
RTO.
Why having an exception here for theADD_ADDR? I mean: I'm OK with
the
idea of doubling the RTO, but only if there is a reason behind that
:)Quoted 5 lines
How about this one (answered by DeepSeek, an AI from China):
/*
*icsk->icsk_rtois not used directly since it's not
* updated frequently enough.What about: (...) since it might not have been updated recently.
Quoted 6 lines
* Setting
subflow_rtoto twice of RTO provides a safety
* buffer for RTT estimation variance, kernel RTO update
* delays, and a full transmission retry cycle while
* minimizing false failures.A bit long and not clear (typical AI, making commit messages too verbose, it almost force me to use one to give me the summary :) )
What about:
Setting this RTO to twice of TCP one, not to be too agressive for this
notification that might take longer to process than "simple" data.Still, is
icsk_rtonot already "long" enough by default? I mean:
when I
see your description in the related packetdrill PR [1] where the delay
is supposed to be 0.01 sec, but "subflow_rto" (or was itisck_rto?) is
0.65 sec, that's quite different, and half of it would be more than
enough :)Quoted 6 lines
Quoted 2 lines
But maybe that's different when it happens later during the connection?
Why wouldn't
icsk_rtonot be good enough ? We just need a reasonable
number here.I didn't have time to investigate in between, but is
icsk_rtonot updated only when needed (@Geliang: did you check?). So could it be possible that at the beginning of the connection, the value is not correct?If yes, why not using
__tcp_set_rto()?Quoted 4 lines
And, I must say that 2 x
__tcp_set_rto()seems a bit arbitrary. Let's just useicsk_rto. Does it really matter that it is
not updated very frequently ?Maybe yes.
@Geliang: but before sending a new version, please check my comment:
https://github.com/multipath-tcp/packetdrill/pull/166
Especially this part:
Quoted 9 lines
0.65s looks a lot to me: The 3rd ACK at line 15 should be injected with a delay of 0.01s.
What are the values of
icsk->icsk_rto,tp->srtt_us,tp->rttvar_us,__tcp_set_rto(tp)and the one you set for the timer when scheduling each retransmissions and when running this test?(for the rto value in jiffies, please convert them to usec like the others →
jiffies_to_usecs())
Quoted 6 lines
Sharing such info when running the packetdrill tests using a non debug
kernel (and ideally also sharing the output of packetdrill, when running
it with the extra -vvvv), would help better understand the situation here.
Here is the debug patch based on v4:
net/mptcp/pm.c +0 -0
@@ -271,6 +271,12 @@ static unsigned int
mptcp_adjust_add_addr_timeout(struct mptcp_sock *msk)
mptcp_for_each_subflow(msk, subflow) {
struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
struct inet_connection_sock *icsk = inet_csk(ssk);+ struct tcp_sock *tp = tcp_sk(ssk);
+
+ pr_info("icsk->icsk_rto=%uus, __tcp_set_rto(tp)=%uus,
tp->srtt_us=%uus, tp->rttvar_us=%uus",
+ jiffies_to_usecs(icsk->icsk_rto), + jiffies_to_usecs(__tcp_set_rto(tp)), + tp->srtt_us, tp->rttvar_us);
if (icsk->icsk_rto > max)
max = icsk->icsk_rto;
@@ -319,9 +325,14 @@ static void mptcp_pm_add_timer(struct timer_list
*timer)
entry->retrans_times++;
}- if (entry->retrans_times < ADD_ADDR_RETRANS_MAX)
+ if (entry->retrans_times < ADD_ADDR_RETRANS_MAX) {
+ pr_info("retrans_times=%u add_addr_timeout=%uus\n",
+ entry->retrans_times,
+ jiffies_to_usecs(timeout << entry-
Quoted 1 line
retrans_times));
+
sk_reset_timer(sk, timer,
jiffies + (timeout << entry-
Quoted 1 line
retrans_times));
+ }
spin_unlock_bh(&msk->pm.lock);
And here is the log:
OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_server.pkt
(ipv6)]
[ 374.343386] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83640us, tp->rttvar_us=200000us [ 374.343396] MPTCP: retrans_times=1 add_addr_timeout=422000us [ 374.343695] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83520us, tp->rttvar_us=200000us [ 374.343696] MPTCP: retrans_times=1 add_addr_timeout=422000us [ 374.351277] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=81064us, tp->rttvar_us=200000us [ 374.351280] MPTCP: retrans_times=1 add_addr_timeout=422000us [ 374.359279] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83416us, tp->rttvar_us=200000us [ 374.359280] MPTCP: retrans_times=1 add_addr_timeout=422000us [ 374.362808] MPTCP: icsk->icsk_rto=433000us,
__tcp_set_rto(tp)=433000us, tp->srtt_us=1121560us, tp-
Quoted 1 line
rttvar_us=291850us
[ 374.362907] MPTCP: icsk->icsk_rto=312000us,
__tcp_set_rto(tp)=312000us, tp->srtt_us=827589us, tp-
Quoted 1 line
rttvar_us=207860us
[ 374.367273] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=80728us, tp->rttvar_us=200000us [ 374.367324] MPTCP: retrans_times=1 add_addr_timeout=422000us
FAIL [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_retry_plain.pkt (ipv4)]
stdout:
stderr:
add_addr_retry_plain.pkt:22: error handling packet: timing error:
expected outbound packet at 1.164616 sec but happened at 0.382298 sec; tolerance 0.400000 sec
script packet: 1.164616 . 1:1(0) ack 1 <add_address address_id: 1 ipv4: 192.168.0.2 hmac: 18072054855919674796> actual packet: 0.382298 . 1:1(0) ack 1 win 1050 <add_address address_id: 1 ipv4: 192.168.0.2 hmac: 18072054855919674796>
Full log is attached.
In this test, __tcp_set_rto(tp) is the same as icsk->icsk_rto.
Thanks,
-Geliang
Quoted 3 lines
Cheers,
Matt
root@mptcpdev:/opt/packetdrill/gtests/net# ./packetdrill/run_all.py -lv mptcp/add_addr/ [ 164.023274] MPTCP: icsk->icsk_rto=310000us, __tcp_set_rto(tp)=310000us, tp->srtt_us=826456us, tp->rttvar_us=206614us [ 372.324724] MPTCP: icsk->icsk_rto=212000us, __tcp_set_rto(tp)=212000us, tp->srtt_us=95520us, tp->rttvar_us=200000us [ 372.334676] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=81552us, tp->rttvar_us=200000us [ 372.349761] MPTCP: icsk->icsk_rto=212000us, __tcp_set_rto(tp)=212000us, tp->srtt_us=90816us, tp->rttvar_us=200000us OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_client.pkt (ipv4)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_port_client.pkt (ipv4)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_server.pkt (ipv6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_client.pkt (ipv4-mapped-v6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_port_client.pkt (ipv4-mapped-v6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_server.pkt (ipv4)] [ 373.062029] MPTCP: icsk->icsk_rto=212000us, __tcp_set_rto(tp)=212000us, tp->srtt_us=93024us, tp->rttvar_us=200000us OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_client.pkt (ipv6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_server.pkt (ipv4-mapped-v6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr4_port_client.pkt (ipv6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_client.pkt (ipv4)] [ 373.303826] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=85528us, tp->rttvar_us=200000us OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_client.pkt (ipv4-mapped-v6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_port_client.pkt (ipv4)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_port_client.pkt (ipv6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_server.pkt (ipv4-mapped-v6)] [ 373.726341] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=82856us, tp->rttvar_us=200000us OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_client.pkt (ipv6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_server.pkt (ipv4)] [ 373.977508] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=81896us, tp->rttvar_us=200000us [ 374.125663] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83520us, tp->rttvar_us=200000us [ 374.125737] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83640us, tp->rttvar_us=200000us [ 374.132917] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=81064us, tp->rttvar_us=200000us [ 374.142526] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83416us, tp->rttvar_us=200000us OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_port_client.pkt (ipv4-mapped-v6)] [ 374.154510] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=80728us, tp->rttvar_us=200000us [ 374.191280] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=81896us, tp->rttvar_us=200000us [ 374.191537] MPTCP: retrans_times=1 add_addr_timeout=422000us OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr6_server.pkt (ipv6)] [ 374.343386] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83640us, tp->rttvar_us=200000us [ 374.343396] MPTCP: retrans_times=1 add_addr_timeout=422000us [ 374.343695] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83520us, tp->rttvar_us=200000us [ 374.343696] MPTCP: retrans_times=1 add_addr_timeout=422000us [ 374.351277] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=81064us, tp->rttvar_us=200000us [ 374.351280] MPTCP: retrans_times=1 add_addr_timeout=422000us [ 374.359279] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83416us, tp->rttvar_us=200000us [ 374.359280] MPTCP: retrans_times=1 add_addr_timeout=422000us [ 374.362808] MPTCP: icsk->icsk_rto=433000us, __tcp_set_rto(tp)=433000us, tp->srtt_us=1121560us, tp->rttvar_us=291850us [ 374.362907] MPTCP: icsk->icsk_rto=312000us, __tcp_set_rto(tp)=312000us, tp->srtt_us=827589us, tp->rttvar_us=207860us [ 374.367273] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=80728us, tp->rttvar_us=200000us [ 374.367324] MPTCP: retrans_times=1 add_addr_timeout=422000us FAIL [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_retry_plain.pkt (ipv4)] stdout: stderr: add_addr_retry_plain.pkt:22: error handling packet: timing error: expected outbound packet at 1.164616 sec but happened at 0.382298 sec; tolerance 0.400000 sec script packet: 1.164616 . 1:1(0) ack 1 <add_address address_id: 1 ipv4: 192.168.0.2 hmac: 18072054855919674796> actual packet: 0.382298 . 1:1(0) ack 1 win 1050 <add_address address_id: 1 ipv4: 192.168.0.2 hmac: 18072054855919674796> FAIL [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_retry_plain.pkt (ipv4-mapped-v6)] stdout: stderr: add_addr_retry_plain.pkt:22: error handling packet: timing error: expected outbound packet at 1.164602 sec but happened at 0.382412 sec; tolerance 0.400000 sec script packet: 1.164602 . 1:1(0) ack 1 <add_address address_id: 1 ipv4: 192.168.0.2 hmac: 10396528243396168920> actual packet: 0.382412 . 1:1(0) ack 1 win 1050 <add_address address_id: 1 ipv4: 192.168.0.2 hmac: 10396528243396168920> FAIL [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_retry_plain.pkt (ipv6)] stdout: stderr: add_addr_retry_plain.pkt:22: error handling packet: timing error: expected outbound packet at 1.083163 sec but happened at 0.295888 sec; tolerance 0.400000 sec script packet: 1.083163 . 1:1(0) ack 1 <add_address address_id: 1 ipv6: fd3d:a0b:17d6::2 hmac: 15336863693204847186> actual packet: 0.295888 . 1:1(0) ack 1 win 1050 <add_address address_id: 1 ipv6: fd3d:a0b:17d6::2 hmac: 15336863693204847186> [ 374.615405] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=81896us, tp->rttvar_us=200000us [ 374.615414] MPTCP: retrans_times=2 add_addr_timeout=844000us [ 374.693298] MPTCP: icsk->icsk_rto=302000us, __tcp_set_rto(tp)=302000us, tp->srtt_us=803324us, tp->rttvar_us=200912us [ 374.767297] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83520us, tp->rttvar_us=200000us [ 374.767489] MPTCP: retrans_times=2 add_addr_timeout=844000us [ 374.767658] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83640us, tp->rttvar_us=200000us [ 374.767659] MPTCP: retrans_times=2 add_addr_timeout=844000us [ 374.775398] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=81064us, tp->rttvar_us=200000us [ 374.775406] MPTCP: retrans_times=2 add_addr_timeout=844000us [ 374.783281] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83416us, tp->rttvar_us=200000us [ 374.783283] MPTCP: retrans_times=2 add_addr_timeout=844000us [ 374.791282] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=80728us, tp->rttvar_us=200000us [ 374.791286] MPTCP: retrans_times=2 add_addr_timeout=844000us OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_echo_new_sf.pkt (ipv4)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_echo_new_sf.pkt (ipv4-mapped-v6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_echo_new_sf.pkt (ipv6)] [ 375.671518] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=80728us, tp->rttvar_us=200000us [ 375.671535] MPTCP: icsk->icsk_rto=211000us, __tcp_set_rto(tp)=211000us, tp->srtt_us=83640us, tp->rttvar_us=200000us OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_retry_errors.pkt (ipv4)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_retry_errors.pkt (ipv4-mapped-v6)] OK [/opt/packetdrill/gtests/net/mptcp/add_addr/add_addr_retry_errors.pkt (ipv6)] Ran 27 tests: 24 passing, 3 failing, 0 timed out (5.41 sec): mptcp/add_addr/
Matthieu Baerts
maintainerRe: [PATCH mptcp-next v3 2/3] mptcp: make ADD_ADDR retransmission timeout adaptive
Hi Geliang,
Quoted 2 lines · Geliang Tang
Quoted 1 line · Matthieu Baerts
On 11/08/2025 17:34, Christoph Paasch wrote:
(...)
Quoted 34 lines
Quoted 34 lines
Quoted 3 lines
Why wouldn't
icsk_rtonot be good enough ? We just need a reasonable
number here.I didn't have time to investigate in between, but is
icsk_rtonot updated only when needed (@Geliang: did you check?). So could it be possible that at the beginning of the connection, the value is not correct?If yes, why not using
__tcp_set_rto()?Quoted 4 lines
And, I must say that 2 x
__tcp_set_rto()seems a bit arbitrary. Let's just useicsk_rto. Does it really matter that it is
not updated very frequently ?Maybe yes.
@Geliang: but before sending a new version, please check my comment:
https://github.com/multipath-tcp/packetdrill/pull/166
Especially this part:
Quoted 9 lines
0.65s looks a lot to me: The 3rd ACK at line 15 should be injected with a delay of 0.01s.
What are the values of
icsk->icsk_rto,tp->srtt_us,tp->rttvar_us,__tcp_set_rto(tp)and the one you set for the timer when scheduling each retransmissions and when running this test?(for the rto value in jiffies, please convert them to usec like the others →
jiffies_to_usecs())
Quoted 8 lines
Quoted 6 lines
Sharing such info when running the packetdrill tests using a non debug
kernel (and ideally also sharing the output of packetdrill, when running
it with the extra -vvvv), would help better understand the situation here.Here is the debug patch based on v4:
(...)
Thank you for having checked! I suggest discussing the logs on the packetdrill ticket where I just asked you to check when not all the packetdrill tests are executed in parallel, certainly taking all CPU resources:
https://github.com/multipath-tcp/packetdrill/pull/166
Quoted 1 line
In this test,
__tcp_set_rto(tp)is the same asicsk->icsk_rto.
That's interesting. Then maybe we can simply use icsk_rto if that's easier.
Cheers,
Matt
Signature · 1 line
Sponsored by the NGI0 Core fund.
Geliang Tang
reviewerPATCH v3 3/3 selftests: mptcp: remove add_addr_timeout settings
In reply to Geliang TangFrom: Geliang Tang <[email protected]>
Now that add_addr_timeout can be dynamically adjusted, there is no need to set specific timeout values in the mptcp_join.sh tests. This patch removes the explicit sysctl settings for net.mptcp.add_addr_timeout from the test scripts.
The change simplifies the test setup and ensures the tests work with the default or dynamically adjusted timeout values.
--- tools/testing/selftests/net/mptcp/mptcp_join.sh | 3 --- 1 file changed, 3 deletions(-)
tools/testing/selftests/net/mptcp/mptcp_join.sh +0 -3
@@ -347,8 +347,6 @@ reset_with_add_addr_timeout() tables="${ip6tables}" fi - ip netns exec $ns1 sysctl -q net.mptcp.add_addr_timeout=1 - if ! ip netns exec $ns2 $tables -A OUTPUT -p tcp \ -m tcp --tcp-option 30 \ -m bpf --bytecode \
@@ -2183,7 +2181,6 @@ signal_address_tests() pm_nl_add_endpoint $ns2 10.0.4.2 flags signal # the peer could possibly miss some addr notification, allow retransmission - ip netns exec $ns1 sysctl -q net.mptcp.add_addr_timeout=1 speed=slow \ run_tests $ns1 $ns2 10.0.1.1
Signature · 1 line
2.48.1
Hi Geliang,
Thank you for your modifications, that's great!
Our CI did some validations and here is its report:
- KVM Validation: normal: Unstable: 2 failed test(s): packetdrill_add_addr selftest_mptcp_connect 🔴
- KVM Validation: debug: Unstable: 1 failed test(s): selftest_mptcp_connect 🔴
- 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/16778676332
Initiator: Matthieu Baerts (NGI0)
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/287779d926ff
Patchwork: https://patchwork.kernel.org/project/mptcp/list/?series=988606
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-normalFor 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)
Matthieu Baerts
maintainerHi Geliang,
Quoted 7 lines · 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: Unstable: 2 failed test(s):
packetdrill_add_addrselftest_mptcp_connect🔴
Don't forget to update the packetdrill tests.
Also, if you are still using mptcp-upstream-virtme-docker, please make sure you are using a recent version where the tolerances are now only increased in debug mode.
Cheers,
Matt
Signature · 1 line
Sponsored by the NGI0 Core fund.
Geliang Tang
reviewerHi Matt,
Quoted 12 lines · Matthieu Baerts
Hi Geliang,
Quoted 8 lines · 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: Unstable: 2 failed test(s):
packetdrill_add_addrselftest_mptcp_connect🔴
Quoted 6 lines
Don't forget to update the packetdrill tests.
Also, if you are still using mptcp-upstream-virtme-docker, please make
sure you are using a recent version where the tolerances are now only increased in debug mode.
Here it is:
https://github.com/multipath-tcp/packetdrill/pull/166
Thanks,
-Geliang
Quoted 3 lines
Cheers,
Matt
Matthieu Baerts
maintainerHi Geliang,
Quoted 17 lines · Geliang Tang
Hi Matt,
Quoted 14 lines · Matthieu Baerts
Hi Geliang,
Quoted 8 lines · 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: Unstable: 2 failed test(s):
packetdrill_add_addrselftest_mptcp_connect🔴Don't forget to update the packetdrill tests.
Quoted 8 lines
Quoted 4 lines
Also, if you are still using mptcp-upstream-virtme-docker, please make
sure you are using a recent version where the tolerances are now only increased in debug mode.Here it is:
Thank you!
I will check ASAP (but I'm still travelling).
Cheers,
Matt
Signature · 1 line
Sponsored by the NGI0 Core fund.