From: HAT <ha...@fa...> - 2007-03-11 16:34:19
|
Hi, >What you're saying about netatalk's signatures is correct. If you >don't set it, the signature it uses is based on the IP address being >used. I think OS X generates a random number which is a little more >useful. In AFP >= 2.x, signature is 16 bytes. Using asip-status.pl, I have already examined the MacOS's signatures on one machine. MacOS 9.2.2 00 00 17 76 00 00 17 76 00 00 17 76 00 00 17 76 Unknown pattern - 4 times MacOS X 10.1.5 and 10.2 31 39 32 2e 31 36 38 2e 31 2e 31 35 00 69 67 2d ASCII code "192.168.1.15<NUL>ig-" leading Bytes is IP address, tailing is unknown MacOS X 10.4 00 03 93 63 ea 36 00 00 00 06 00 00 06 0e 00 00 leading 6 Bytes is MAC address, tailing 10 Bytes is unknown These OS support only AFP 2.x and later. I do netatalk's signature, too. netatalk 2.1dev (Linux Intel 32bit) 01 01 c0 a8 01 01 c0 a8 01 01 c0 a8 01 01 c0 a8 IP address - 4 times netatalk 2.0.3 (Linux Intel 64bit) 01 05 c0 a8 00 00 00 00 01 05 c0 a8 00 00 00 00 IP address - 2 times, because gethostid() is 64bit netatalk 2.0.3 (Solaris9 Intel 32bit) a4 4c a0 06 a4 4c a0 06 a4 4c a0 06 a4 4c a0 06 hostid - 4 times, is not IP address Intel-Solaris hostid is rundom number netatalk support AFP 1.1 and later. >However, what this client is doing is just considering the first 2 >bytes, regardless of what it is given. It could be that the client is >just wrong. > >What is it you're really trying to do, though? Do you have a Mac OS 7 >machine you're using as a client? > >What would be more useful is to get a capture of the traffic between >two Mac OS 7 machines speaking AFP 1.1 to each other. Then we could >guess at some of the protocol going across. My AFP 1.1 client is afpclient command only. I have the Mac OS 9 and later. It is not possible to experiment on it because I do not have Mac OS 7. The "MultipleServers" problem is FAQ. I want to write a patch for unique signature. If AFP 1.1 signature is 2 Bytes, I should devise it. -- HAT |