PMFS and driver model backing devices / device nodes

[Matthew Wilcox]

On Mon, Nov 11, 2013 at 02:42:50PM -0800, Andy Lutomirski wrote:
> Do you know if anyone has given any thought to what a backing device
> should look like?  ISTM that passing a physaddr as a mount option is
> messy and that it would be much nicer if you could mount some kind of
> device node.  As an added benefit, this would let things like
> populating the image on startup be done separately from the underlying
> filesystem.
> 
> XIP block devices are somewhat close to being the right thing.

Yes, that's exactly the plan :-)

See, for example, this thread:
https://lkml.org/lkml/2013/9/4/555