[PATCH v3 mptcp-net 0/5] mptcp: fix fallback-related races

[Paolo Abeni <pabeni@redhat.com>]

This series contains 3 fixes somewhat related to various races we have
while handling fallback and 2 small follow-up likely more suited for
net-next.

The root cause of the issues addressed here is that the check for
"we can fallback to tcp now" and the related action are not atomic. That
also applies to fallback due to MP_FAIL - where the window race is even
wider.

Address the issue introducing an additional spinlock to bundle together
all the relevant events, as per patch 1 and 2.

Note that mptcp_disconnect() unconditionally
clears the fallback status (zeroing msk->flags) but don't tuch the
`allows_infinite_fallback` flag. Such issue is addressed in patch 3.

Patch 4 cleans up a bit the fallback code, introducing specific MIB for
each FB reason, and patch 5 drops the, hopefully now redundant
pr_fallback().
---
v2 -> v3:
 - mptcp_do_fallback  -> mptcp_try_fallback
 - refactored patch 3/5
 - changed mibs names, increment fail only when the protocol mandate it
 - fix W=1 warn in patch 5/5

Paolo Abeni (5):
  mptcp: make fallback action and fallback decision atomic
  mptcp: plug races between subflow fail and subflow creation
  mptcp: reset fallback status gracefully at disconnect() time
  mptcp: track fallbacks accurately via mibs
  mptcp: remove pr_fallback()

 net/mptcp/ctrl.c     |  4 +-
 net/mptcp/mib.c      |  5 +++
 net/mptcp/mib.h      |  7 ++++
 net/mptcp/options.c  |  6 ++-
 net/mptcp/pm.c       |  8 +++-
 net/mptcp/protocol.c | 97 ++++++++++++++++++++++++++++++++++----------
 net/mptcp/protocol.h | 35 ++++++++--------
 net/mptcp/subflow.c  | 40 ++++++++++--------
 8 files changed, 140 insertions(+), 62 deletions(-)

-- 
2.50.0