Re: 230-objrmap fixes for 2.6.3-mjb2

[Andrea Arcangeli <andrea@suse.de>]

On Wed, Mar 03, 2004 at 02:58:20AM -0800, Andrew Morton wrote:
> Andrea Arcangeli <andrea@suse.de> wrote:
> >
> > --- sles-objrmap/mm/rmap.c.~1~	2004-03-03 06:45:38.995594456 +0100
> >  +++ sles-objrmap/mm/rmap.c	2004-03-03 07:01:39.200621104 +0100
> >  @@ -470,7 +470,7 @@ try_to_unmap_obj_one(struct vm_area_stru
> >   	if (!pte)
> >   		goto out;
> >   
> >  -	if (vma->vm_flags & VM_LOCKED) {
> >  +	if (vma->vm_flags & (VM_LOCKED|VM_RESERVED)) {
> >   		ret =  SWAP_FAIL;
> >   		goto out_unmap;
> 
> I keep on wanting to put that in there too.  But pages in a VM_RESERVED vma
> should not find their way onto the LRU.  Maybe we should be checking for
> that in do_no_page().

you're right, so we can turn this into a BUG_ON, I prefer it here, since
this is really the place that VM_RESERVED is mean to guard against (plus
the vma merging).

thanks for the explanation.