[PATCH V3 0/4] 9p: Convert to the new mount API

[Eric Sandeen <sandeen@redhat.com>]

This is another updated attempt to convert 9p to the new mount API. 9p
is one of the last conversions needed, possibly because it is one of the
trickier ones!

This has had only light testing. I confirmed that with the diod server,
both of these mount successfully:

# mount -t 9p -o aname=/tmp/9,version=9p2000.L,uname=fsgqa,access=user,trans=tcp 127.0.0.1 /mnt

and

# mount.diod localhost:/tmp/9 /mnt

(The latter passes trans=fd under the covers)

I have not been able to test other transports, or exhaustively test
functionality of all mount options.

Changes from V1 to V2:
        Address "make W=1" warnings from kernel test robot, comments from
        dhowells, and some kernel-doc comments for changed arguments.

Changes from V2 to V3:

Patch 1:
        None
Patch 2:
        None
Patch 3:
        Change to not re-use v9fs_session_info and p9_client in the
        v9fs_context structure. Instead, new structures p9_client_opts and
        p9_session_opts are introduced. This avoids confusion about what is
        used only for option parsing, and what is used for the actual
        mounted instance.
Patch 4:
        Allows unknown mount options as prior code did. I noticed that
        mount.diod passes a "rwdepth" option by default, which was ignored
        previously but breaks mount with strict unknown option rejection.

        Adjust variable names in v9fs_parse_param, v9fs_apply_options,
        and v9fs_init_fs_context to reflect new context structure and
        to clarify difference.

        Limit the msize option to INT_MAX as the old code did.

        Remove redundant v9fs_set_super function, set_anon_super is enough.

        Properly extract v9fs_seesion_info from sb->s_fs_info, not
        the fc context, in v9fs_fill_super. (thanks eadavis@qq.com)

        Remove unneeded fc arg from v9fs_fill_super.