Re: [PATCH v3 mptcp-next 2/3] mptcp: more careful RM_ADDR generation

[Paolo Abeni <pabeni@redhat.com>]

On Fri, 2022-02-11 at 15:10 -0800, Mat Martineau wrote:
> On Thu, 10 Feb 2022, Paolo Abeni wrote:
> 
> > In some edge scenarios, an MPTCP subflows can use a local address
> > mapped by a "dummy" endpoint created by the in kernel path manager.
> > 
> > When such endpoint is deleted, the in kernel PM sends a RM_ADDR MPTCP
> > suboption. That is somewhat unexpected, as an MPTCP listener will keep
> > accepting incoming subflows targeting such address and the unexpected
> > options can confuse some self-tests.
> > 
> > Be more conservative about RM_ADDR generation: do it only if the
> > relevant endpoint has either the SIGNAL or SUBFLOW flag - it's not a
> > dummy one.
> > 
> > Signed-off-by: Paolo Abeni <pabeni@redhat.com>
> > ---
> > include/uapi/linux/mptcp.h |  1 +
> > net/mptcp/pm_netlink.c     | 25 +++++++++++++++++--------
> > 2 files changed, 18 insertions(+), 8 deletions(-)
> > 
> > diff --git a/include/uapi/linux/mptcp.h b/include/uapi/linux/mptcp.h
> > index f106a3941cdf..34ca8c04f64e 100644
> > --- a/include/uapi/linux/mptcp.h
> > +++ b/include/uapi/linux/mptcp.h
> > @@ -81,6 +81,7 @@ enum {
> > #define MPTCP_PM_ADDR_FLAG_SUBFLOW			(1 << 1)
> > #define MPTCP_PM_ADDR_FLAG_BACKUP			(1 << 2)
> > #define MPTCP_PM_ADDR_FLAG_FULLMESH			(1 << 3)
> > +#define MPTCP_PM_ADDR_FLAG_DUMMY			(1 << 4)
> 
> Since this is a public API, "DUMMY" might be a confusing / ambiguous name. 
> MPTCP_PM_ADDR_FLAG_IMPLICIT_ENDPOINT or MPTCP_PM_ADDR_FLAG_UNADVERTISED 
> maybe? (open to other ideas of course)

I think "IMPLICIT" is the better option, as it's both unadvertised, not
used for subflow.
> 
> It looks like these dummy/implicit records stay around until a flush 
> happens. What if there's a request to advertise an address that has had a 
> dummy created already? mptcp_pm_nl_append_new_local_addr() would consider 
> that a duplicate and reject it, but replacing the dummy record with a real 
> one would be better.

Agreed. I'll do that in the next iteration.

/P