more strict KPI for vm_pager_get_pages()

[Gleb Smirnoff]

On Wed, May 06, 2015 at 11:59:51AM -0500, Alan Cox wrote:
A> > Pagers are responsible to update the array of pages in
A> > case if they replace pages in an object. All array entries
A> > must be valid, if pager returns VM_PAGER_OK.
A> >
A> > Note: the only pager that replaces pages is sg_pager, and
A> > it does that correctly.
A> >
A> 
A> No, look again, e.g., device_pager.

Yes, now I see it, it goes trough cdev op.

But it appears also to work correctly - it updates the page in the array.

-- 
Totus tuus, Glebius.