[ofw] patch 2/2 Add support for RDMAoEth to the low level driver

[James Yang]

It's a field inside cm req path record. This path information is passed over wire during cm hand shaking, and remote will decide to use gid or local lid to communicate.

You're right on the second one. I think I misinterpreted your comments in my earlier email.

Thanks,
James

-----Original Message-----
From: Hefty, Sean [mailto:sean.hefty at intel.com] 
Sent: Thursday, December 03, 2009 3:45 PM
To: James Yang; Tzachi Dar; ofw at lists.openfabrics.org
Subject: RE: [ofw] patch 2/2 Add support for RDMAoEth to the low level driver

>> >+ // make this same as Linux code: p_path->hop_flow_raw >= 2
>> means grh present.
>> >+ if(cl_hton32(p_path->hop_flow_raw) >= 2)  p_av->attr.grh_valid = 
>> >+ 1;
>>
>> Please submit this as a separate patch, and it would be best to 
>> decide on a single way to set grh_valid.
>>
>>
>
>This is required otherwise remote cm (like linux) will not use gid for 
>the communication and failed cm.

I don't follow this.  This is setting a local variable, not something that goes on the wire from what I can tell.  How does this change impact the remote side?

>This remote gid is used to let local knows where to send info. Just 
>like an ip address.
>It's not IB environment. For simple test program like pingpong, specify 
>gid here makes life easy. Otherwise how do you find out remote gid?

The same way the remote LID and QPN are determined.  The test programs determine the values using local calls and exchange those with the remote side out of band.  This makes the programs simpler to use.