[MPTCP] Re: [MPTCP][PATCH v3 mptcp-next 0/5] move to next addr when timeout

[MPTCP] Re: [MPTCP][PATCH v3 mptcp-next 0/5] move to next addr when timeout

[Mat Martineau]

[-- Attachment #1: Type: text/plain, Size: 1259 bytes --]

On Wed, 24 Feb 2021, Geliang Tang wrote:

> v3:
> - invoke lookup_subflow_by_daddr in mptcp_pm_nl_add_addr_received.
> - add a new patch "drop argument port from mptcp_pm_announce_addr".
>
> v2:
> - invoke lookup_subflow_by_daddr in mptcp_pm_add_addr_received, not in
> mptcp_pm_nl_add_addr_received.
> - add a sanity check on cfg_do_w.
>
> Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/151
>
> Geliang Tang (5):
>  mptcp: drop argument port from mptcp_pm_announce_addr
>  mptcp: skip connecting the connected address
>  mptcp: move to next addr when timeout
>  selftests: mptcp: add cfg_do_w for cfg_remove
>  selftests: mptcp: timeout testcases for multi addresses
>
> net/mptcp/pm.c                                |  6 ++--
> net/mptcp/pm_netlink.c                        | 35 +++++++++++++++----
> net/mptcp/protocol.h                          |  2 +-
> .../selftests/net/mptcp/mptcp_connect.c       | 10 ++++--
> .../testing/selftests/net/mptcp/mptcp_join.sh | 26 ++++++++++++--
> 5 files changed, 64 insertions(+), 15 deletions(-)
>
> -- 
> 2.29.2

Thanks Geliang, I think v3 is ready for the export branch.


Reviewed-by: Mat Martineau <mathew.j.martineau(a)linux.intel.com>

--
Mat Martineau
Intel

[MPTCP] Re: [MPTCP][PATCH v3 mptcp-next 0/5] move to next addr when timeout

[Geliang Tang]

[-- Attachment #1: Type: text/plain, Size: 1881 bytes --]

Hi Matt,

Matthieu Baerts <matthieu.baerts(a)tessares.net> 于2021年2月26日周五 下午3:58写道：
>
> Hi Geliang,
>
> On 26/02/2021 04:42, Geliang Tang wrote:
> > Matthieu Baerts <matthieu.baerts(a)tessares.net> 于2021年2月25日周四 下午11:01写道：
> >> Tests + export have been queued!
> >
> > I'm sorry to bring you extra work and trouble. But my patches on the
> > export/20210225T204843 branch still need to be adjusted the order.
> >
> > This series "move to next addr when timeout" should be put after the
> > series "move to next addr and testcases" like this:
>
> I can sure change the order but do we need to change it because some
> patches depends on others that are set after?
>
> I should certainly explain why it is in this order:
>
> - Initially I put the "rm list" series
>
> - Then I added "mptcp: free resources when the port number is
> mismatched" and "mptcp: drop unused subflow in
> mptcp_pm_subflow_established" because they were looking like fixes for
> net-next
>
> - Then I had to put "move to next addr when timeout" series before "rm
> list" because some patches had to be inserted before "mptcp: drop unused
> subflow in mptcp_pm_subflow_established"
>
> But I can understand if the dependences are not OK. If it is, that's not
> an issue to change the order. If not, I guess we can leave it like that.
> We will end up sending everything at the next merge window anyway.

The dependence is fine, only the "move to next addr" series (issue 151) had
been cut off into two parts by the "rm list" series. It looks a little
strange. I thought some selftests would fail, but I just tested them, they
all work well. So if we can accept the cut-off series, let's leave it
there. :)

Thanks.

-Geliang

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

[MPTCP] Re: [MPTCP][PATCH v3 mptcp-next 0/5] move to next addr when timeout

[Matthieu Baerts]

[-- Attachment #1: Type: text/plain, Size: 1381 bytes --]

Hi Geliang,

On 26/02/2021 04:42, Geliang Tang wrote:
> Matthieu Baerts <matthieu.baerts(a)tessares.net> 于2021年2月25日周四 下午11:01写道：
>> Tests + export have been queued!
> 
> I'm sorry to bring you extra work and trouble. But my patches on the
> export/20210225T204843 branch still need to be adjusted the order.
> 
> This series "move to next addr when timeout" should be put after the
> series "move to next addr and testcases" like this:

I can sure change the order but do we need to change it because some 
patches depends on others that are set after?

I should certainly explain why it is in this order:

- Initially I put the "rm list" series

- Then I added "mptcp: free resources when the port number is 
mismatched" and "mptcp: drop unused subflow in 
mptcp_pm_subflow_established" because they were looking like fixes for 
net-next

- Then I had to put "move to next addr when timeout" series before "rm 
list" because some patches had to be inserted before "mptcp: drop unused 
subflow in mptcp_pm_subflow_established"

But I can understand if the dependences are not OK. If it is, that's not 
an issue to change the order. If not, I guess we can leave it like that. 
We will end up sending everything at the next merge window anyway.

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

[MPTCP] Re: [MPTCP][PATCH v3 mptcp-next 0/5] move to next addr when timeout

[Geliang Tang]

[-- Attachment #1: Type: text/plain, Size: 2880 bytes --]

Hi Matt,

Matthieu Baerts <matthieu.baerts(a)tessares.net> 于2021年2月25日周四 下午11:01写道：
>
> Hi Geliang, Mat,
>
> On 24/02/2021 04:56, Geliang Tang wrote:
> > v3:
> >   - invoke lookup_subflow_by_daddr in mptcp_pm_nl_add_addr_received.
> >   - add a new patch "drop argument port from mptcp_pm_announce_addr".
> >
> > v2:
> >   - invoke lookup_subflow_by_daddr in mptcp_pm_add_addr_received, not in
> >   mptcp_pm_nl_add_addr_received.
> >   - add a sanity check on cfg_do_w.
> >
> > Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/151
>
> Thank you for the patches and the reviews!
> I just added them in our tree with Mat's RvB tag:
>
> Patches 1/5 and 2/5 have been added before "mptcp: drop unused subflow
> in mptcp_pm_subflow_established":
>
> - d63677a633cf: mptcp: drop argument port from mptcp_pm_announce_addr
> - c7ef42ed2451: mptcp: skip connecting the connected address
> - a681ac4046d5: conflict in t/mptcp-add-rm_list_tx-in-mptcp_pm_data
> - Results: de59595c18b3..52c9bb446e5d
>
> The rest just after the mentioned patch:
>
> - 1d7d152ed66f: mptcp: move to next addr when timeout
> - 9f635adaa8c3: selftests: mptcp: add cfg_do_w for cfg_remove
> - da596d8d5312: selftests: mptcp: timeout testcases for multi addresses
> - Results: 52c9bb446e5d..8a82f6b7cf01
>
> Tests + export have been queued!

I'm sorry to bring you extra work and trouble. But my patches on the
export/20210225T204843 branch still need to be adjusted the order.

This series "move to next addr when timeout" should be put after the
series "move to next addr and testcases" like this:

  mptcp: free resources when the port number is mismatched
  mptcp: add rm_list in mptcp_out_options
  mptcp: add rm_list_tx in mptcp_pm_data
  mptcp: add rm_list in mptcp_options_received
  mptcp: add rm_list_rx in mptcp_pm_data
  mptcp: remove multi addresses in PM
  mptcp: remove multi subflows in PM
  mptcp: remove multi addresses and subflows in PM
  mptcp: remove a list of addrs when flushing
  selftests: mptcp: add invert argument for chk_rm_nr
  selftests: mptcp: set addr id for removing testcases
  selftests: mptcp: add testcases for removing addrs
  mptcp: export lookup_anno_list_by_saddr
  mptcp: move to next addr when subflow creation fail
  selftests: mptcp: signal addresses testcases
  mptcp: drop argument port from mptcp_pm_announce_addr
  mptcp: skip connecting the connected address
  mptcp: drop unused subflow in mptcp_pm_subflow_established
  mptcp: move to next addr when timeout
  selftests: mptcp: add cfg_do_w for cfg_remove
  selftests: mptcp: timeout testcases for multi addresses

Could you please help me to re-apply them in this order?

Thanks very much.

-Geliang




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

[MPTCP] Re: [MPTCP][PATCH v3 mptcp-next 0/5] move to next addr when timeout

[Matthieu Baerts]

[-- Attachment #1: Type: text/plain, Size: 1296 bytes --]

Hi Geliang, Mat,

On 24/02/2021 04:56, Geliang Tang wrote:
> v3:
>   - invoke lookup_subflow_by_daddr in mptcp_pm_nl_add_addr_received.
>   - add a new patch "drop argument port from mptcp_pm_announce_addr".
> 
> v2:
>   - invoke lookup_subflow_by_daddr in mptcp_pm_add_addr_received, not in
>   mptcp_pm_nl_add_addr_received.
>   - add a sanity check on cfg_do_w.
> 
> Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/151

Thank you for the patches and the reviews!
I just added them in our tree with Mat's RvB tag:

Patches 1/5 and 2/5 have been added before "mptcp: drop unused subflow 
in mptcp_pm_subflow_established":

- d63677a633cf: mptcp: drop argument port from mptcp_pm_announce_addr
- c7ef42ed2451: mptcp: skip connecting the connected address
- a681ac4046d5: conflict in t/mptcp-add-rm_list_tx-in-mptcp_pm_data
- Results: de59595c18b3..52c9bb446e5d

The rest just after the mentioned patch:

- 1d7d152ed66f: mptcp: move to next addr when timeout
- 9f635adaa8c3: selftests: mptcp: add cfg_do_w for cfg_remove
- da596d8d5312: selftests: mptcp: timeout testcases for multi addresses
- Results: 52c9bb446e5d..8a82f6b7cf01

Tests + export have been queued!

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

[MPTCP] Re: [MPTCP][PATCH v3 mptcp-next 0/5] move to next addr when timeout

[Geliang Tang]

[-- Attachment #1: Type: text/plain, Size: 2052 bytes --]

Hi Matt,

Mat Martineau <mathew.j.martineau(a)linux.intel.com> 于2021年2月25日周四 上午10:23写道：
>
> On Wed, 24 Feb 2021, Geliang Tang wrote:
>
> > v3:
> > - invoke lookup_subflow_by_daddr in mptcp_pm_nl_add_addr_received.
> > - add a new patch "drop argument port from mptcp_pm_announce_addr".
> >
> > v2:
> > - invoke lookup_subflow_by_daddr in mptcp_pm_add_addr_received, not in
> > mptcp_pm_nl_add_addr_received.
> > - add a sanity check on cfg_do_w.
> >
> > Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/151
> >
> > Geliang Tang (5):
> >  mptcp: drop argument port from mptcp_pm_announce_addr
> >  mptcp: skip connecting the connected address
> >  mptcp: move to next addr when timeout
> >  selftests: mptcp: add cfg_do_w for cfg_remove
> >  selftests: mptcp: timeout testcases for multi addresses

This patch "mptcp: drop unused subflow in mptcp_pm_subflow_established" in
the export branch should be put in this patchset like this:

  mptcp: drop argument port from mptcp_pm_announce_addr
  mptcp: skip connecting the connected address
  mptcp: drop unused subflow in mptcp_pm_subflow_established
  mptcp: move to next addr when timeout
  selftests: mptcp: add cfg_do_w for cfg_remove
  selftests: mptcp: timeout testcases for multi addresses

But I sent it alone a few days earlier. Could you please help me to insert
it back into this patchset when merging them?

Thanks.

-Geliang

> >
> > net/mptcp/pm.c                                |  6 ++--
> > net/mptcp/pm_netlink.c                        | 35 +++++++++++++++----
> > net/mptcp/protocol.h                          |  2 +-
> > .../selftests/net/mptcp/mptcp_connect.c       | 10 ++++--
> > .../testing/selftests/net/mptcp/mptcp_join.sh | 26 ++++++++++++--
> > 5 files changed, 64 insertions(+), 15 deletions(-)
> >
> > --
> > 2.29.2
>
> Thanks Geliang, I think v3 is ready for the export branch.
>
>
> Reviewed-by: Mat Martineau <mathew.j.martineau(a)linux.intel.com>
>
> --
> Mat Martineau
> Intel