lkml.org 
[lkml]   [2006]   [Apr]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [RFC][PATCH] swsusp: support creating bigger images
Date
Hi,

On Wednesday 26 April 2006 00:24, Nigel Cunningham wrote:
> On Wednesday 26 April 2006 08:21, Rafael J. Wysocki wrote:
> > On Tuesday 25 April 2006 23:18, Nigel Cunningham wrote:
> > > On Wednesday 26 April 2006 07:12, Rafael J. Wysocki wrote:
> > > > On Tuesday 25 April 2006 22:32, Pavel Machek wrote:
> > > > > > > > -unsigned int count_data_pages(void)
> > > > > > > >
> > > > > > > > +/**
> > > > > > > > + * need_to_copy - determine if a page needs to be copied
> > > > > > > > before saving. + * Returns false if the page can be saved
> > > > > > > > without copying. + */
> > > > > > > > +
> > > > > > > > +static int need_to_copy(struct page *page)
> > > > > > > > +{
> > > > > > > > + if (!PageLRU(page) || PageCompound(page))
> > > > > > > > + return 1;
> > > > > > > > + if (page_mapped(page))
> > > > > > > > + return page_mapped_by_current(page);
> > > > > > > > +
> > > > > > > > + return 1;
> > > > > > > > +}
> > > > > > >
> > > > > > > I'd much rather VM internal type stuff get moved *out* of
> > > > > > > kernel/power :(
> > > > > >
> > > > > > Well, I kind of agree, but I don't know where to place it under
> > > > > > mm/.
> > > > > >
> > > > > > > It needs more comments too. Also, how important is it for the
> > > > > > > page to be off the LRU?
> > > > > >
> > > > > > Hm, I'm not sure if that's what you're asking about, but the pages
> > > > > > off the LRU are handled in a usual way, ie. copied when
> > > > > > snapshotting the system. The pages _on_ the LRU may be included in
> > > > > > the snapshot image without copying, but I require them additionally
> > > > > > to be (a) mapped by someone and (b) not mapped by the current task.
> > > > >
> > > > > Why do you _want_ them mapped by someone?
> > > >
> > > > Because this means they belong to a task that is frozen and won't touch
> > > > them (of course unless it's us). The kernel has no reason to access
> > > > them either (even after we resume devices) except for reclaiming, but
> > > > that's handled explicitly. Thus it's safe to include them in the image
> > > > without copying.
> > > >
> > > > As I said before, I think the page cache pages may be treated this way
> > > > too. It probably applies to all of the LRU pages, but there may be some
> > > > corner cases. The mapped pages are just easy to single out.
> > >
> > > It does apply to all of the LRU pages. This is what I've been doing for
> > > years now. The only corner case I've come across is XFS. It still wants
> > > to write data even when there's nothing to do and it's threads are frozen
> > > (IIRC - haven't looked at it for a while). I got around that by freezing
> > > bdevs when freezing processes.
> >
> > This means if we freeze bdevs, we'll be able to save all of the LRU pages,
> > except for the pages mapped by the current task, without copying. I think
> > we can try to do this, but we'll need a patch to freeze bdevs for this
> > purpose. ;-)
>
> Isn't that a coincidence? I just happen to have the code right here! Not in a
> patch form that you'd want though - do you have any refrigerator changes in
> mm at the moment, or can I use vanilla 2.6.17-rc2 process.c?

I think you can use the 2.6.17-rc2 one safely.

Greetings,
Rafael

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2006-04-26 00:46    [W:0.197 / U:0.048 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site