Comment 5 for bug 1786902

If I omit ddb457c6993b "net/smc: Replace ib_query_gid with rdma_get_gid_attr" all the rest of the commits listed in comment #2 cherry pick cleanly. That commit is later reverted, though intermediate changes make it not a direct revert, but that does make think we can probably just omit it. If I do that I'm left with a net/smc that basically only differs from Linus' master in that it's lacking b4c296f9c964 "RDMA/smc: Replace ib_query_gid with rdma_get_gid_attr" (aside from an include missing in a couple of files). Does this sound like a reasonable approach?