diff mbox series

[mptcp-next] mptcp: init autotune state also in simultaneous connect case.

Message ID 20200622135737.8217-1-fw@strlen.de
State Accepted, archived
Commit e6df694a4180
Delegated to: Matthieu Baerts
Headers show
Series [mptcp-next] mptcp: init autotune state also in simultaneous connect case. | expand

Commit Message

Florian Westphal June 22, 2020, 1:57 p.m. UTC 
This is what Christoph suggested needs to be added to avoid
div0 in case of simultaneous connect.

Squashto: mptcp: add receive buffer auto-tuning
---
 net/mptcp/subflow.c | 1 +
 1 file changed, 1 insertion(+)

Comments

Paolo Abeni June 30, 2020, 10:54 a.m. UTC | #1 
On Mon, 2020-06-22 at 15:57 +0200, Florian Westphal wrote:
> This is what Christoph suggested needs to be added to avoid
> div0 in case of simultaneous connect.
> 
> Squashto: mptcp: add receive buffer auto-tuning
> ---
>  net/mptcp/subflow.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
> index 0f0fa1ba57a8..e1e19c76e267 100644
> --- a/net/mptcp/subflow.c
> +++ b/net/mptcp/subflow.c
> @@ -1120,6 +1120,7 @@ static void subflow_state_change(struct sock *sk)
>  
>  	if (subflow_simultaneous_connect(sk)) {
>  		mptcp_do_fallback(sk);
> +		mptcp_rcv_space_init(mptcp_sk(parent), sk);
>  		pr_fallback(mptcp_sk(parent));
>  		subflow->conn_finished = 1;
>  		if (inet_sk_state_load(parent) == TCP_SYN_SENT) {

LGTM, and I think Christoph already acked it.

Thanks Florian,

/P
Mat Martineau June 30, 2020, 6:44 p.m. UTC | #2 
On Tue, 30 Jun 2020, Paolo Abeni wrote:

> On Mon, 2020-06-22 at 15:57 +0200, Florian Westphal wrote:
>> This is what Christoph suggested needs to be added to avoid
>> div0 in case of simultaneous connect.
>>
>> Squashto: mptcp: add receive buffer auto-tuning
>> ---
>>  net/mptcp/subflow.c | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
>> index 0f0fa1ba57a8..e1e19c76e267 100644
>> --- a/net/mptcp/subflow.c
>> +++ b/net/mptcp/subflow.c
>> @@ -1120,6 +1120,7 @@ static void subflow_state_change(struct sock *sk)
>>
>>  	if (subflow_simultaneous_connect(sk)) {
>>  		mptcp_do_fallback(sk);
>> +		mptcp_rcv_space_init(mptcp_sk(parent), sk);
>>  		pr_fallback(mptcp_sk(parent));
>>  		subflow->conn_finished = 1;
>>  		if (inet_sk_state_load(parent) == TCP_SYN_SENT) {
>
> LGTM, and I think Christoph already acked it.

Yes, definitely good to squash this.

--
Mat Martineau
Intel
Matthieu Baerts July 1, 2020, 11:42 a.m. UTC | #3 
Hi Florian, Mat, Paolo,

On 30/06/2020 20:44, Mat Martineau wrote:
> On Tue, 30 Jun 2020, Paolo Abeni wrote:
> 
>> On Mon, 2020-06-22 at 15:57 +0200, Florian Westphal wrote:
>>> This is what Christoph suggested needs to be added to avoid
>>> div0 in case of simultaneous connect.
>>>
>>> Squashto: mptcp: add receive buffer auto-tuning
>>
>> LGTM, and I think Christoph already acked it.
> 
> Yes, definitely good to squash this.

Thank you for the patch and the review! Just applied:

- e6df694a4180: "squashed" in "mptcp: add receive buffer auto-tuning"
- 8de6e2516b12..d8ab2db3ac90: result

Tests and export will be started soon.

Cheers,
Matt
diff mbox series

Patch

diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 0f0fa1ba57a8..e1e19c76e267 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1120,6 +1120,7 @@  static void subflow_state_change(struct sock *sk)
 
 	if (subflow_simultaneous_connect(sk)) {
 		mptcp_do_fallback(sk);
+		mptcp_rcv_space_init(mptcp_sk(parent), sk);
 		pr_fallback(mptcp_sk(parent));
 		subflow->conn_finished = 1;
 		if (inet_sk_state_load(parent) == TCP_SYN_SENT) {