[MPTCP] [Weekly meetings] MoM - 5th of November 2020

[MPTCP] [Weekly meetings] MoM - 5th of November 2020

[Matthieu Baerts]

[-- Attachment #1: Type: text/plain, Size: 9235 bytes --]

Hello everyone,

Last Thursday, we had our 123rd meeting with Mat and Ossama (Intel OTC), 
Christoph (Apple), Davide, Paolo and Florian (RedHat) 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, Mat Martineau):

1393323  [net-next,v2,7/7] selftests: mptcp: add ADD_ADDR timeout test case
1393322  [net-next,v2,6/7] docs: networking: mptcp: Add MPTCP sysctl entries
1393321  [net-next,v2,5/7] mptcp: add a new sysctl add_addr_timeout
1393318  [net-next,v2,4/7] mptcp: split mptcp_clean_una function
1393320  [net-next,v2,3/7] tcp: propagate MPTCP skb extensions on xmit 
splits
1393319  [net-next,v2,2/7] mptcp: use _fast lock version in 
__mptcp_move_skbs
1393317  [net-next,v2,1/7] mptcp: adjust mptcp receive buffer limit if 
subflow...

1392003  [net] mptcp: token: fix unititialized variable


     our repo (by: Geliang Tang):

1390711  [v2,mptcp-next] Squash to "mptcp: change add_addr_signal type"



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: Florian Westphal, Geliang Tang, Mat Martineau, Paolo 
Abeni):

1370700: RFC: [RFC,2/4] tcp: move selected mptcp helpers to tcp.h/mptcp.h
1370701: RFC: [RFC,3/4] mptcp: add mptcp reset option support
1370702: RFC: [RFC,4/4] tcp: parse tcp options contained in reset packets:
     - WIP, same as last week

1375893: RFC: [RFC,mptpcp-next] mptcp: add ooo prune support:
     - WIP, same as last week

1387845: RFC: MPTCP stream performances:
     - can be drop

1390270: RFC: [RFC] mptcp: refine MPTCP-level ack scheduling:
     - can be drop, v1 sent

1392576: Changes Requested: [mptcp-next] docs: networking: mptcp: Add 
MPTCP sysctl entries:
     - applied

1394437: New: [net-next] Squash-to "mptcp: refactor shutdown and close":
     - depends on some local changes Paolo has on his side
     - v2 needed to have something building

1394439: New: [v1] mptcp: refine MPTCP-level ack scheduling
     -  can be reviewed: this has a fix from an issue mentioned earlier

1394788: New: [v3,mptcp-next,01/12] mptcp: unify ADD_ADDR and echo 
suboptions writing
1394789: New: [v3,mptcp-next,02/12] mptcp: unify ADD_ADDR and ADD_ADDR6 
suboptions writing
1394790: New: [v3,mptcp-next,03/12] mptcp: add port support for ADD_ADDR 
suboption writing
1394791: New: [v3,mptcp-next,04/12] mptcp: use adding up size to get 
ADD_ADDR length
1394792: New: [v3,mptcp-next,05/12] mptcp: add the outgoing ADD_ADDR 
port support
1394793: New: [v3,mptcp-next,06/12] mptcp: send out dedicated packet for 
ADD_ADDR using port
1394794: New: [v3,mptcp-next,07/12] mptcp: add port parameter for 
mptcp_pm_announce_addr
1394795: New: [v3,mptcp-next,08/12] mptcp: print out port and ahmac when 
receiving ADD_ADDR
1394796: New: [v3,mptcp-next,09/12] mptcp: deal with 
MPTCP_PM_ADDR_ATTR_PORT in PM netlink
1394797: New: [v3,mptcp-next,10/12] selftests: mptcp: add port argument 
for pm_nl_ctl
1394798: New: [v3,mptcp-next,11/12] mptcp: add the mib for ADD_ADDR with 
port
1394799: New: [v3,mptcp-next,12/12] selftests: mptcp: add testcases for 
ADD_ADDR with port:
     - for the listen part, we need to bind on another port
     - should be fine if we make sure the binaries are the same, from 
the same user (maybe problematic) and in the same NS
     - but can we easily check that? Is it safe?
     - idea: have mptcpd creating this listen socket (and announce the 
port)?
     - how can we ensure only trusted application can open these other 
ports?
     - goal (security): avoid having another app opening a port to 
bypass some rules limiting an app on a specific port
     - maybe safer: the application should tell the kernel 
(netlink/sockopt) to accept MP_JOIN from another ports (and maybe 
interface/IP) and ask to announce ADD_ADDR over another port

1394896: New: [RFC,mptcp-next,1/6] mptcp: add the outgoing MP_PRIO support
1394898: New: [RFC,mptcp-next,2/6] mptcp: add the incoming MP_PRIO support
1394899: New: [RFC,mptcp-next,3/6] mptcp: add prio_changed flag
1394900: New: [RFC,mptcp-next,4/6] mptcp: send out MP_PRIO before RM_ADDR
1394901: New: [RFC,mptcp-next,5/6] mptcp: add the mib for MP_PRIO
1394902: New: [RFC,mptcp-next,6/6] selftests: mptcp: add MP_PRIO check 
in chk_rm_nr

Note:
     - Florian has some patches related to MP_FASTCLOSE
     - he will send them after the meeting



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


     Recently opened (latest from last week: 104)

   106  [syzkaller] BUG: Bad page state [bug] [syzkaller]:
       - might be linked to 104 where there is a reproducer
       - no reproducer for 106 but the trace looks the same as the 104

   105  warning in `mptcp_reset_timer()` with `mptcp_connect.sh -m mmap` 
[bug]:
       - can be linked to 70
       - *Matth:* to check


     Bugs (opened, flagged as "bug" and assigned)

    94  Packetdrill: after a received DATA_FIN, no new packets can be 
treated [bug] @dcaratti:
        - WIP

    85  Packetdrill: multiple timeout reported by the CI [bug] @matttbe:
        - for the moment, only issue with mp_join (timeout after 2 minutes)


     Bugs (opened and flagged as "bug" and not assigned)

   106  [syzkaller] BUG: Bad page state [bug] [syzkaller]
   105  warning in `mptcp_reset_timer()` with `mptcp_connect.sh -m mmap` 
[bug]
   104  [syzkaller] general protection fault in skb_release_data [bug] 
[syzkaller]
   103  [syzkaller] WARNING in inet_csk_listen_stop [bug] [syzkaller]
    99  simult_flows selftest is unstable: remaining sockets in 
TIME-WAIT state [bug]
    70  [syzkaller] WARNING in mptcp_reset_timer [bug] [syzkaller]
    65  clearing properly the status in listen() [bug]
    56  msk connection state set without msk lock [bug]


     In Progress (opened and assigned)

   96  Python: add support for IPPROTO_MPTCP [enhancement] @matttbe
   76  [gs]etsockopt per subflow: BPF [enhancement] @matttbe
   54  ADD_ADDR: ports support [enhancement] @geliangtang
   51  MP_PRIO support [enhancement] @geliangtang

   49: MP_FASTCLOSE
       - Florian is working on that

   43  [syzkaller] Change syzkaller to exercise MPTCP inet_diag 
interface [enhancement] [syzkaller] @cpaasch


     Recently closed (since last week)

None.



Warnings (WARN_ON):
     - We should be careful about them, especially for the benign ones
     - It is costly (especially if you have them in a loop)
     - Some people use a sysctl to panic on a warning



Refactoring:
     - Paolo is working on (yet another) refactoring :-)
     - → getting rid of a workqueue for the receiving part



Commits to send to Netdev:
     - Paolo would like to some patches to be sent
     - he will send a proposition later



Davide on Fedora:
     - goal: send mptcpd to have it in Fedora 34
     - spec are ready but need to be validated
     - https://gitlab.com/dcaratti/mptcpd/-/blob/main/mptcpd.spec
     - not sure if other distro are packaging it
     - there is a git hook that triggers a build when we commit on the 
gitlab repo
     - ideally, it could be nice to have a githook from the official 
mptcpd github repo to trigger a build in case of new tag
     - Could be good to also propose packages for other distro like 
Debian (Ubuntu) → Matth? :)



FYI: Current Roadmap:
     - Bugs: https://github.com/multipath-tcp/mptcp_net-next/projects/2
     - Current merge window (5.11): 
https://github.com/multipath-tcp/mptcp_net-next/projects/6
     - For later: https://github.com/multipath-tcp/mptcp_net-next/projects/4



Extra tests:
     - news about Syzkaller? (Christoph):
         - See Github ↑

     - news about interop with mptcp.org? (Christoph):
         - Had some issues last week, will retry

     - news about Intel's kbuild? (Mat):
         - haven't receive any new build report
         - Mat will investigate why the builds have stopped

     - packetdrill (Davide):
         - See Github ↑

     - CI (Matth):
         - tests looks a bit more stable but still some are failing 
(packetdrill, see above)



CGroup BPF hook on socket creation:
     - Mat looked at that
     - it would be a bit different from the others (we don't have the 
socket, we then cannot retrieve the cgroup from the "usual" way)



Next meeting:
     - We propose to have the next meeting on Thursday, the 12th of 
November.
     - Usual UTC time: 16:00 UTC (8am PST, 5pm CET, Midnight CST)
     - Still open to everyone!
     - https://annuel2.framapad.org/p/mptcp_upstreaming_20201112



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

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