[asterisk-dev] Skinny protocol messages with recent Cisco phones-

[Dan Austin]

Damien wrote:
> Dan Austin wrote:
>> I don't have wireshark, so I am open to suggestions.
>> My sniffer is reporting it as 'extra bytes'.  At least
>> one of structures in the current code is using :
>>
>> uint32_t space;
>>
>> To generate 4 bytes of padding, so I extended the concept.
>> I have a patch that I am cleaning up right now, and I
>> can/will open a bug with it attached.
>>
>> My 7940 is still functional after these changes, so the
>> older phones do not care if the packet is padded, but the
>> new ones choke if it is not.  Based on that observation, I
>> would tend to think that this approach is valid not only
>> for trunk, but also branches/1.4 since the code tries to
>> support the new devices.
>>
>> An alternate approach is to duplicate all of the structures
>> and have versions for old and new phones and make the general
>> code aware of the difference so we send old style packets
>> to legacy models and the new style for the current models (yuck).
>>
>> Dan
> I think the older devices are fairly tolerant to different size 
> packets. 
> Some of the structures in chan_skinny are still not complete, 
> meaning they are shorter than anticipated. If the phones needed 
> all of the data, no devices would be working on the current code
> (example, CALL_INFO_MESSAGE is still missing some stuff at the 
> end).
Actually that structure looks good.  Asterisk was missing the
padding, but according the protocol decode my sniffer offered up,
all of the fields were accurately identified, present and the
correct sizes.  One interesting thing I noted is that Asterisk
is padding with nulls, where as CCM appears to pad with random
data (or at lease gibberish to me and my sniffer)

> It's probably reasonable to assume that these devices accept 
> longer packets and just grab the stuff they want. Your initial 
> am/pm patch wouldn't have worked as you added another byte to the 
> data structure, but it did, so I think it's safe to say that the 
> old devices just ignore the extra data.

> Damien


> PS I don't like the idea of duplicating all the structures.
Me either.  It should not be needed, but it exists as an option.

Dan

Oh and who ever feed the listserv its Wheaties today deserves
a pay raise, this is the most responsive I've ever seen it...