[openib-general] [PATCH] fmr pool: remove unnecessary pointer dereference

[Or Gerlitz]

On 7/10/06, Roland Dreier <rdreier at cisco.com> wrote:
> Yes, looks like a good idea since the underlying FMR API doesn't
> change the IOVA.

So this means that a ULP can not map on the same time these two page sets:
 <A,B> and <A,B,C> suggesting the verbs layer to have A being the IOVA
at the HCA IOMMU (eg MPT/MTT in the mellanox case), and getting some
A* to be used for the second map?

Such a mapping might be possible eg under zero copy (direct IO) test
doing two reads on parallel.

Or.