[PATCH v5 05/15] mm/frame-vector: Use FOLL_LONGTERM

[Christoph Hellwig]

On Wed, Nov 04, 2020 at 04:54:19PM +0100, Daniel Vetter wrote:
> I don't really have a box here, but dma_mmap_attrs() and friends to
> mmap dma_alloc_coherent memory is set up as VM_IO | VM_PFNMAP (it's
> actually enforced since underneath it uses remap_pfn_range), and
> usually (except if it's pre-cma carveout) that's just normal struct
> page backed memory. Sometimes from a cma region (so will be caught by
> the cma page check), but if you have an iommu to make it
> device-contiguous, that's not needed.

dma_mmap_* memory may or may not be page backed, but it absolutely
must not be resolved by get_user_pages and friends as it is special.
So yes, not being able to get a struct page back from such an mmap is
a feature.