Previously we were completely over allocating, fix this
by actually implementing the size calculation.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>