Roland Dreier wrote:
> --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
> + dev->ibdev.local_dma_lkey = 0;
why this specific initialization but not in nes or amso1100? I think
the device structure is zeroed out on allocation anyway.
I wasn't sure if I should assume that. Plus setting it explicitly shows
that we're supporting the feature.
WRT amso and nes, they don't even set the capability bit in the device
structure at registration time. They only return when querying the
device. So I punted. :)
> __be64 node_guid;
> u8 node_type;
> u8 phys_port_cnt;
> + u32 local_dma_lkey;
It's a little silly of me to mention this, given how few of these
structures are used, but I think it would be nicer to put local_dma_lkey
after node_guid to avoid leaving padding in the structure.
- R.
Can do.
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
