Re: [RFC v1 1/4] dma-mapping: introduce the DMA_ATTR_MAY_SLEEP attribute

[Christoph Hellwig]

On Tue, Mar 28, 2023 at 09:21:10AM +0200, Petr Tesarik wrote:
> The full series in my local tree added it to the implementation of
> DRM_IOCTL_PRIME_FD_TO_HANDLE:

Umm, an all these are callers that absolutely never should even
end up in swiotlb.  If we have large buffers allocated by media
subsystems, we need to make sure they are fully addressable.