Re: LDL Driver: cannot send to arbitrary LLC (802.2) DSAP

Dave,

Sorry if I've come across as rude and (or) stupid. In the last month
I've had problems with large software companies not implementing
standard interfaces to, well, standards. And it's starting to gnaw
at me. Sorry to be sending it

The LiS software is an absolute godsend, and aside from a couple of
hacks that I had to make in my 2.6 kernel headers (multiple structure
declarations), it has built and runs superbly.

With the exception of this SSAP==DSAP filtering; with this switched off it
works very similarly to the AIX and Solaris drivers. So I was surprised
that code had been ADDED (i.e. not a case of laziness or sloppiness) to do
the filtering.

For now, I'm just going to hack the code out - I see no need to go to the
raw layer. But is there a "better", more "correct" way of dealing with this
so other users of ldl can benefit?

I didn't recognise the word "transparent", since all the way through the
ldl driver you use the word "raw" more often.

Good work, keep it up :-)

Regards

Tim.

Dave Grothe wrote:
> At 10:45 AM 9/16/2004, Tim Brown wrote:
> 
>> Dave,
>>
>> What do you mean by: "The Transparent Mode"?
> 
> 
> I don't have time to give you the cookbook, but if you look at the ldl 
> driver (and, who knows, perhaps even its documentation) there are 
> different modes of operation that are controlled by how you do the 
> dl_attach.  One is a transparent mode in which your driver gets to see 
> all of the incoming Ethernet packets.
> 
> -- Dave
> 
> 
>> tim
>>
>> Dave Grothe wrote:
>>
>>> I have been using the transparent mode in order to avoid this kind of 
>>> thing.  My driver does its own filtering of addresses and SAPs.
>>> -- Dave
>>> At 09:33 AM 9/16/2004, Tim Brown wrote:
>>>
>>>> I am porting some network multiplexor software which uses multiplexes
>>>> messages between hosts and processes on hosts using 802.2 DSAP/SSAP
>>>> addressing.
>>>>
>>>> I bind onto a DSAP, allowing me to listen for Ethernet frames addressed
>>>> to me. Within those packets is the return address for me to send my 
>>>> replies
>>>> to. This may or may not be the Ethernet address of the sender.
>>>>
>>>> I then wish to send a frame to the destination which may have an 
>>>> arbitrary
>>>> DSAP. However, in eth_8022_mkhdr() in drivers/str/linux/ldl.c line:1704
>>>>
>>>>         if (dsap != dl->sap->sap.sap[0]) {
>>>>                 struct sap *sap;
>>>>
>>>>                 for (sap = dl->subs; sap; sap = sap->next_sap)
>>>>                         if (dsap == sap->sap.sap[0])
>>>>                                 break;
>>>>                 if (sap == NULL)
>>>>                         return 0;
>>>>         }
>>>>
>>>> i.e. if you are not bound to "dsap", or have an additional binding
>>>>      to "dsap", then the message isn't sent.
>>>>
>>>> I am using LiS 2.17.U.
>>>>
>>>> I don't believe this behaviour is correct, since it is inconsistent
>>>> with the Solaris and AIX DLPI drivers, as well as the understanding
>>>> of most people here of how DSAP/SSAP addressing works; i.e. you bind
>>>> onto and listen on a DSAP (or multiple DSAPs), but any process on any
>>>> station can transmit to any of the available SSAPs.
>>>>
>>>> Therefore, I don't believe there is any reason for this code. Is there?
>>>>
>>>> Regards,
>>>>
>>>> Tim
>>>>
>> Linux-streams mailing list
>> [email protected]
>> http://gsyc.escet.urjc.es/mailman/listinfo/linux-streams

-- 
Tim Brown <[email protected]> |            City Computing Limited |
T: +44 20 8770 2110               |      City House, Sutton Park Road |
F: +44 20 8770 2130               |       Sutton, Surrey, SM1 2AE, GB |
BEAUTY: What's in your eye when you have a bee in your hand.__________/