video: host: tegra: avoid linear mapping to iova space

video: host: tegra: avoid linear mapping to iova space

Avoid dma_linear_map() call as virtualization can cause iova address
space to be split among the guest OSes.

So, map the carveout to iova space dynamically and pass the iova start
and size as parameters to the ucode.

Bug 200022149

Change-Id: Ie89d650c8b5959fc79fd11391b95c15ea97df9ee
Signed-off-by: Sri Krishna chowdary <schowdary@nvidia.com>
Reviewed-on: http://git-master/r/563040
GVS: Gerrit_Virtual_Submit
Reviewed-by: Sharath Sarangpur <ssarangpur@nvidia.com>
Reviewed-by: Krishna Reddy <vdumpa@nvidia.com>