There are two scenarios that we will restore deleted records. The first is
when device down and up(or unmap/remap). In this scenario the new filter
mode is same with previous one. Because we get it from in_dev->mc_list and
we do not touch it during device down and up.
The other scenario is when a new socket join a group which was just delete
and not finish sending status reports. In this scenario, we should use the
current filter mode instead of restore old one. Here are 4 cases in total.
old_socket        new_socket       before_fix       after_fix
  IN(A)             IN(A)           ALLOW(A)         ALLOW(A)
  IN(A)             EX( )           TO_IN( )         TO_EX( )
  EX( )             IN(A)           TO_EX( )         ALLOW(A)
  EX( )             EX( )           TO_EX( )         TO_EX( )
Fixes: 24803f38a5c0b (igmp: do not remove igmp souce list info when set link down)
Fixes: 1666d49e1d416 (mld: do not remove mld souce list info when set link down)
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
        spin_lock_bh(&im->lock);
        if (pmc) {
                im->interface = pmc->interface;
-               im->sfmode = pmc->sfmode;
-               if (pmc->sfmode == MCAST_INCLUDE) {
+               if (im->sfmode == MCAST_INCLUDE) {
                        im->tomb = pmc->tomb;
                        im->sources = pmc->sources;
                        for (psf = im->sources; psf; psf = psf->sf_next)
 
        spin_lock_bh(&im->mca_lock);
        if (pmc) {
                im->idev = pmc->idev;
-               im->mca_sfmode = pmc->mca_sfmode;
-               if (pmc->mca_sfmode == MCAST_INCLUDE) {
+               if (im->mca_sfmode == MCAST_INCLUDE) {
                        im->mca_tomb = pmc->mca_tomb;
                        im->mca_sources = pmc->mca_sources;
                        for (psf = im->mca_sources; psf; psf = psf->sf_next)