Re: [PATCH] swiotlb: sync buffer when mapping FROM_DEVICE

[Robin Murphy]

On 22/05/2019 13:32, Christoph Hellwig wrote:
> I'm a little worried about this.  While it looks functionally correct
> we have surived without it, and doing another copy for every swiotlb
> dma mapping from the device looks extremely painful for the typical use
> cases where we expect the device to transfer the whole mapping.
>
> I'd be tempted to instead properl document the current behavior and
> introduce a new DMA_ATTR_PARTIAL flag to allow for partial mappings.

Would that work out any different from the existing 
DMA_ATTR_SKIP_CPU_SYNC? If drivers are prepared to handle this issue 
from their end, they can already do so for single mappings by using that 
attr along with explicit partial syncs via dma_sync_single(). For 
page/sg mappings we'd still have the problem of identifying what part of 
"partial" actually matters, and probably having to add some additional 
new sync operations to cope.

Robin.