[ofa-general] Re: [PATCH 1/3] mcast loopback block

[Or Gerlitz]

Or Gerlitz wrote:
> This way, each call to the MCG attach verb would be duplicated in the 
> code, since if it fails when the "block loopback" param was set it 
> would now be called again unset, etc. I think this turns to be more 
> complex then doing a query for a cap bit and then setting the block 
> mcast through a QP create flag once. To my taste, the fact that the QP 
> create flag is used by the driver at the time of MCG attach does not 
> make it a more complex implementation.
Hi Roland,

Can you comment on this thread, we want to get the patches into 2.6.27 
and then to libibverbs.

thanks,

Or.