|
-----Original Message-----
???: Peter Kese <peter.kese@xxxxxx>
???: Wensong Zhang <wensong@xxxxxxxxxxxx>
??: linux-virtualserver <linux-virtualserver@xxxxxxxxxxxx>
??: 1999?2?10? 17:53
??: tunneling & masq_app
>
>Hi!
>
>I believe it would be possible to set up an FTP Virtual Server with
>Tunneling. In order to do that, the ip_masq_ftp.c code should be
>partially reimplemented to work only on upstream part of the TCP
>connection.
>
Yeah, it should be possible. But, I think setting up an FTP Virtual
Server doesn't need to reimplement part of the ip_masq_ftp.c code,
because there is no relation, the tunneling host doesn't need to
parse the PASV command in the ftp packets. You can ftp the tunneled
host, can't you? :-)
FTP virtual server with tunneling should work as you can ftp the
tunneled host. But, note that don't insert ip_masq_ftp.o into
your system, otherwise it will corrupt the ftp packet. If you inserted
ip_masq_ftp.o, please remove it and reboot your system, see what
happens.
>I have been playing arround with the kernel source code. To start with,
>I have put in some printk calls into the ip_masq_ftp.c. It appears, the
>ip_masquerading modules do not get used by the tunneling VS patch.
>
>Unlike the original ip_masq_new(), the VS' ip_masq_new_pfvs() function
>does not call the ip_masq_bind_app(), which is required in order to use
>the module. Therefore, when the ip_fw_demasquerade() calls the
>ip_masq_app_pkt_in() function, the later does not find any masq_app
>entry in the masq structure and simply returns.
>
>Did I get it right? Would it help to add an appropriate call to
>ip_masq_bind_app() into the ip_masq_new_pfvs() or am I forwarding a more
>complicated problem.
>
>Cheers,
> Peter
>
|
