Re: msync() bug

[Linus Torvalds]

On Mon, 9 Jul 2001, Hugh Dickins wrote:
>
> As it stands, yes.  But shouldn't there be some kind of VM_ASIFRESERVED
> vma flag to treat all its pages as if they were reserved, for /dev/mem?

Sure, that is doable. We do in fact already have the flag - you could
think of the VM_IO as that kind of flag already. I wouldn't object to that
kind of change in the 2.5.x timeframe. We already have vmscan ignoring
VM_IO objects, we could do the same to copy_mm() and to mmdrop().

		Linus