The code looks fine, but I still very much disagree with the workaround
comments.  The different mask for allocations vs streaming mappings is
exactly how the DMA API is intended to work.

Reply via email to