Some of my patches in ML are conflicted, rebase them into one series.

1-8: some small cleanups, v2
9-10: address #428, add mptcpi_subflows_total
11-17: address #379, #391, userspace pm remove id 0 subflow & address, v11
18-20: address #403, add refcont for address entry
21: add userspace fullmesh tests
22-30: seltests cleanups

Geliang Tang (30):
  mptcp: drop useless ssk in pm_subflow_check_next
  mptcp: use mptcp_check_fallback helper
  mptcp: use mptcp_get_ext helper
  mptcp: use mptcp_wnd_end helper
  mptcp: move sk assignment statement ahead
  mptcp: define more local variables sk
  selftests: mptcp: sockopt: drop mptcp_connect var
  selftests: mptcp: display simult in extra_msg
  mptcp: add mptcpi_subflows_total counter
  selftests: mptcp: add chk_subflows_total helper
  selftests: mptcp: add evts_get_info helper
  selftests: mptcp: update userspace pm test helpers
  selftests: mptcp: userspace pm remove id 0 subflow
  mptcp: userspace pm allow creating id 0 subflow
  selftests: mptcp: userspace pm create id 0 subflow
  mptcp: userspace pm remove id 0 address
  selftests: mptcp: userspace pm remove id 0 address
  mptcp: add userspace_pm_get_entry helper
  mptcp: add userspace pm addr entry refcount
  mptcp: add netlink pm addr entry refcount
  selftests: mptcp: add userspace pm fullmesh tests
  selftests: mptcp: add mptcp_lib_kill_wait
  selftests: mptcp: add mptcp_lib_evts_*
  selftests: mptcp: userspace: print colored results
  selftests: mptcp: add mptcp_lib_verify_listener_events
  selftests: mptcp: add mptcp_lib_is_v6
  selftests: mptcp: add mptcp_lib_get_counter
  selftests: mptcp: add mptcp_lib_make_file
  selftests: mptcp: add mptcp_lib_check_transfer
  selftests: mptcp: add mptcp_lib_wait_local_port_listen

 include/uapi/linux/mptcp.h                    |   1 +
 net/mptcp/options.c                           |   2 +-
 net/mptcp/pm.c                                |   2 +-
 net/mptcp/pm_netlink.c                        |  25 +-
 net/mptcp/pm_userspace.c                      | 139 ++++--
 net/mptcp/protocol.c                          |   6 +-
 net/mptcp/protocol.h                          |  14 +-
 net/mptcp/sockopt.c                           |   4 +-
 tools/testing/selftests/net/mptcp/diag.sh     |  23 +-
 .../selftests/net/mptcp/mptcp_connect.sh      | 108 +---
 .../testing/selftests/net/mptcp/mptcp_join.sh | 460 +++++++++---------
 .../testing/selftests/net/mptcp/mptcp_lib.sh  | 188 +++++++
 .../selftests/net/mptcp/mptcp_sockopt.sh      |  55 +--
 .../selftests/net/mptcp/simult_flows.sh       |  19 +-
 .../selftests/net/mptcp/userspace_pm.sh       | 199 +++-----
 15 files changed, 662 insertions(+), 583 deletions(-)

-- 
2.35.3

Hi Geliang,

On 19/09/2023 06:02, Geliang Tang wrote:
> Some of my patches in ML are conflicted, rebase them into one series.
> 
> 1-8: some small cleanups, v2
> 9-10: address #428, add mptcpi_subflows_total
> 11-17: address #379, #391, userspace pm remove id 0 subflow & address, v11
> 18-20: address #403, add refcont for address entry
> 21: add userspace fullmesh tests
> 22-30: seltests cleanups

Thank you for this series, it is useful (and again sorry for the delay
to review them). I will really try to look at them ASAP, hopefully soon.
I had a very quick look.

Please note that patches 14/30 and 16/30 will need to go to -net instead
of net-next. I guess it's OK if the selftests linked to them go to
net-next because I think they depend on new features you are adding
here. I think it should not be an issue to apply these two patches
before (maybe a small conflict with "mptcp: define more local variables
sk" but easy to fix).

One more thing: we should be able to run the last version of the
selftests on older kernels (e.g. v5.15 and v6.1). I think with your
modifications, we are fine (I didn't test and only quickly looked) but
that's something to keep in mind.

(also, I think we will need to invert patches 10/30 and 11/30 because
patch 10/30 depends on helpers introduced in the next patch if I'm not
mistaken)

Good idea to share code via the lib then having multiple copies!

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