[MPTCP] [Weekly meetings] MoM - 2nd of June 2020

[MPTCP] [Weekly meetings] MoM - 2nd of June 2020

[Matthieu Baerts]

[-- Attachment #1: Type: text/plain, Size: 8934 bytes --]

Hello,

Last Thursday, we had our 106th meeting with Mat and Ossama (Intel OTC), 
Christoph (Apple), Paolo, Davide and Florian (RedHat), Geliang Tang 
(Xiaomi) and myself (Tessares).

Thanks again for this new good meeting!

Here are the minutes of the meeting:



Accepted patches:
     - The list of accepted patches can be seen on PatchWork:
       https://patchwork.ozlabs.org/project/mptcp/list/?state=3

     netdev (if mptcp ML is in cc) (Davide Caratti, Florian Westphal, 
Paolo Abeni):

1320094  [net-next,2/2] mptcp: add receive buffer auto-tuning
1320091  [net-next,1/2] selftests: mptcp: add option to specify size of
1319952  [net-next] mptcp: do nonce initialization at subflow creation t
1319179  [net-next,6/6] mptcp: close poll() races
1319176  [net-next,5/6] mptcp: __mptcp_tcp_fallback() returns a struct s
1319178  [net-next,4/6] mptcp: create first subflow at msk creation time
1319173  [net-next,3/6] mptcp: check for plain TCP sock at accept time
1319181  [net-next,2/6] mptcp: fallback in case of simultaneous connect
1319172  [net-next,1/6] net: mptcp: improve fallback to TCP
1317876  [net-next,v2,4/4] mptcp: introduce token KUNIT self-tests
1317873  [net-next,v2,3/4] mptcp: move crypto test to KUNIT
1317875  [net-next,v2,2/4] mptcp: refactor token container
1317874  [net-next,v2,1/4] mptcp: add __init annotation on setup functio

     our repo (by: Florian Westphal, Paolo Abeni):

1317568  Squash-to: "net: mptcp: improve fallback to TCP"
1317473  [net] mptcp: fix DSS map generation on fin retransmission
1314402  [mptcp-next] mptcp: init autotune state also in simultaneous co



Pending patches:
     - The list of pending patches can be seen on PatchWork:
       https://patchwork.ozlabs.org/project/mptcp/list/?state=*

     netdev (if mptcp ML is in cc) (by: /):

/

     our repo (by: Mat Martineau, Paolo Abeni):

1318177: Changes Requested: [RFC] mptcp: Use full MPTCP-level disconnect 
state machine:
     - Working on a v2 with a split in different commits

1319751: New: [RFC,1/5] mptcp: cleanup subflow_finish_connect()
1319753: New: [RFC,2/5] subflow: explicitly check for plain tcp rsk
1319754: New: [RFC,3/5] subflow: use rsk_ops->send_reset()
1319756: New: [RFC,4/5] subflow: introduce and use mptcp_can_accept_new_
1319755: New: [RFC,5/5] subflow: do not create child subflow for fallba:
     - Series: mptcp: cope better with mp_join storm
     - Link to issue 33 on Github (weighttp-test)
     - should improve the situation there
     - Waiting for review
     - Waiting for tests from Christoph
     - For the patch 2/5, there is a note about SYN cookie, not 
supported with MPTCP (yet)
     - we should in theory not arrive there for the moment.

1321331: New: [v3,1/4] inet_diag: support for wider protocol numbers
1321329: New: [v3,2/4] mptcp: add msk interations helper
1321332: New: [v3,3/4] mptcp: add MPTCP socket diag interface
1321333: New: [v3,4/4] selftests/mptcp: add diag interface tests:
     - token iterator uses RCU to traverse then a lock on each sock
     - faster than spinlock
     - avoid possible deadlock
     - New version waiting for review

1321336: New: [v2,1/2] include: update mptcp uAPI
1321335: New: [v2,2/2] ss: mptcp: add msk diag interface support:
     - Waiting for review



Issues on Github:
     https://github.com/multipath-tcp/mptcp_net-next/issues/

     open: (latest from last week: 39)

   42  remove unneeded branches in critical path [enhancement]:
       - perf related topic, not urgent
       - should not be difficult
       - linked to issue 15

   41  reduce indirect call usage [enhancement]:
       - perf related topic, not urgent
       - should not be difficult

   39  [syzkaller] WARNING in subflow_data_ready (v2.0) 2 @dcaratti
   34  WARNING: Bad mapping: ssn=1 map_seq=498340137 map_data_len=79
   33  weighttp-test on 1KB file with mptcp.org-client in "ndiffports"-mo
   31  Allow MPTCP + SYN_COOKIES [enhancement]
   30  ssh restart does not work
   24  Revisit layout of struct mptcp_subflow_context [enhancement]
   21  sort-out {set,get}sockopt handling [enhancement]

   20  implement msk diag interface [enhancement] @pabeni:
       - should be fixed by patches recently shared

   19  let PM netlink update live sockets on local addresses list change 
[enhancement]
   18  allow 'force to MPTCP' mode [enhancement]

   17  audit 'backup' flag usage  [enhancement]:
       - Paolo is looking at this
       - to be able to send traffic over multiple subflows at the same time

   15  reduce mptcp_out_option struct size [enhancement]
    6  loss and delay without reordering causes very slow transfer
    4  keep a single work struct in mptcp socket [enhancement]

     recently closed:

   40  Blocking accept() does not return



FYI: Current Roadmap:
     - Part 4 (next merge window):
         - Fix bugs reported on Github: 
https://github.com/multipath-tcp/mptcp_net-next/issues/
         - IPv6 - IPv4 mapped support
         - not dropping MPTCP options (ADD_ADDR, etc.)
         - FAST_CLOSE
         - full MPTCP v1 support (reliable add_addr, rm_addr, etc.)
         - after a few attempts of failed MPTCP, we fallback to TCP 
(cache like TFO is doing)
         - PM server (more advanced) [DONE by Paolo]
         - Active backup support [WIP by Paolo]
         - sending "simultaneously" on different subflows (multiple non 
backup subflows) [WIP by Paolo]
         - easily allow other apps to use MPTCP (sysctl, bpf, systemtap, 
LD_PRELOAD → doc?) [WIP Nicolas]
         - Full DATA_FIN support [WIP by Mat]
         - ADD_ADDR for MPTCPv1: echo bit [WIP by Peter]
         - Opti in TCP option structures (unions) [DONE by Paolo]
         - Shared recv window (full support) [DONE by Florian]
     - Part 5 (extra needed for prod):
         - opti/perfs
         - TFO
         - PM netlink per connection
         - PM bpf
         - Scheduler bpf
         - syncookies [WIP Florian]
         - [gs]etsockopt per subflow
         - notify the userspace when a subflow is added/removed → cmsg

*Matth* check how to use Github for the roadmap



Part 4: new features:
     - news about "Full DATA_FIN support"?:
         - RFC has been shared

     - news about "ADD_ADDR for MPTCPv1: echo bit"?:
         - /
         - Peter is still interested by this



Meetings at a different time:
     - one hour earlier: 8am PDT, 5pm CEST, 11pm UTC+8



Packetdrill issues:
     - See ML for more details
     - some are due to "fs: Do not check if there is a fsnotify watcher 
on pseudo inodes"
     - "fsnotify: suppress access/modify events on stream files" 
(temporary fix) is now in export branch
     - Do we still need it?:
         - Paolo found a patch that has been sent to fix the issue: 
b6509f6a8c4313c068c69785c001451415969e
         - *Matth* : we no longer need the patch we have in the export 
branch (once the above patch is available in net-next)

     - We still have another issue reported on IRC:
         - open an issue on Github
         - seems to be present on slow system, with debug
         - *Matth* : check that



Extra tests:
     - news about Syzkaller? (Christoph):
         - Running on latest export branch with all Paolo patches

     - news about interop with mptcp.org? (Christoph):
         - /

     - news about Intel's kbuild? (Mat):
         - same as before:
             - building of the export branch is still done
             - tests are still unstable (not tested)

     - packetdrill (Davide):
         - 2 PR have been accepted:
             - supporting MP_JOIN (Todd) + additional fixes (Todd):
                 - Matth: check if MP_JOIN tests are checked
             - enable MPTCP (sysctl)

     - CI (Matth):
         - focus on .c reproducer (from next week)

     - Coverity (Davide):
         - /



Syn-cookies:
     - Florian is looking at that
     - Working fine on the MP_CAPABLE
     - how to handle MP_JOIN?:
         - when we are full, we drop
         - in mptcp.org, we limit to 32 accept



- mptcp: fix DSS map generation on FIN retransmission:
     - candidate for -net
     - linked to issue 34
     - does not conflicts with others so we can send
     - (for -net, maybe best to send them on a Friday not to have to ask 
for a sync)



Next meeting:
     - We propose to have the next meeting on Thursday, the 9th of July.
     - *NEW* Usual time: 15:00 UTC (8am PDT, 5pm CEST, 11pm UTC+8)
     - Still open to everyone!
     - https://annuel2.framapad.org/p/mptcp_upstreaming_20200709



Feel free to comment on these points and propose new ones for the next
meeting!

Talk to you next week,
Matt
-- 
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net