On Sat, Sep 27, 2014 at 11:15:16PM +0400, Konstantin Khlebnikov wrote:
> This patch fixes all of them.
It seems to have rendered virtio_balloon completely ineffective without
CONFIG_COMPACTION and CONFIG_BALLOON_COMPACTION, even for the case that I'm
expanding the memory available to my VM.
Was this intended? Should Kconfig be updated so that VIRTIO_BALLOON depends on
BALLOON_COMPACTION now?
On Tue, Oct 28, 2014 at 9:47 PM, Matt Mullins <[email protected]> wrote:
> On Sat, Sep 27, 2014 at 11:15:16PM +0400, Konstantin Khlebnikov wrote:
>> This patch fixes all of them.
>
> It seems to have rendered virtio_balloon completely ineffective without
> CONFIG_COMPACTION and CONFIG_BALLOON_COMPACTION, even for the case that I'm
> expanding the memory available to my VM.
What do you mean by ineffective?
That it cannot handle fragmentation? I saw that without compaction
ballooning works even better:
it allocates pages without GFP_MOVABLE and buddy allocator returns
much less scattered pages.
>
> Was this intended? Should Kconfig be updated so that VIRTIO_BALLOON depends on
> BALLOON_COMPACTION now?
Nope, this is independent feature.
On Tue, Oct 28, 2014 at 11:56:15PM +0400, Konstantin Khlebnikov wrote:
> On Tue, Oct 28, 2014 at 9:47 PM, Matt Mullins <[email protected]> wrote:
> > On Sat, Sep 27, 2014 at 11:15:16PM +0400, Konstantin Khlebnikov wrote:
> >> This patch fixes all of them.
> >
> > It seems to have rendered virtio_balloon completely ineffective without
> > CONFIG_COMPACTION and CONFIG_BALLOON_COMPACTION, even for the case that I'm
> > expanding the memory available to my VM.
>
> What do you mean by ineffective?
Ballooning does not change the amount of memory available at all. Bisecting
the change got me back to this patch.
I'm currently using 3.18-rc2 and running in qemu-kvm-1.6.2-9.fc20.x86_64 via
libvirt-daemon-1.1.3.6-1.fc20.x86_64.
I'll attach my .config; enabling COMPACTION and BALLOON_COMPACTION makes
virtio-ballooning work again.
> That it cannot handle fragmentation? I saw that without compaction
> ballooning works even better:
> it allocates pages without GFP_MOVABLE and buddy allocator returns
> much less scattered pages.
>
> >
> > Was this intended? Should Kconfig be updated so that VIRTIO_BALLOON depends on
> > BALLOON_COMPACTION now?
>
> Nope, this is independent feature.
Good to know, thanks.
Fix for commit d6d86c0a7f8ddc5b38cf089222cb1d9540762dc2
("mm/balloon_compaction: redesign ballooned pages management").
If CONFIG_BALLOON_COMPACTION=n balloon_page_insert() does not link
pages with balloon and doesn't set PagePrivate flag, as a result
balloon_page_dequeue cannot get any pages because it thinks that
all of them are isolated. Without balloon compaction nobody can
isolate ballooned pages, it's safe to remove this check.
Signed-off-by: Konstantin Khlebnikov <[email protected]>
Reported-by: Matt Mullins <[email protected]>
Cc: Stable <[email protected]> (v3.17)
---
mm/balloon_compaction.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/mm/balloon_compaction.c b/mm/balloon_compaction.c
index b3cbe19..fcad832 100644
--- a/mm/balloon_compaction.c
+++ b/mm/balloon_compaction.c
@@ -68,11 +68,13 @@ struct page *balloon_page_dequeue(struct balloon_dev_info *b_dev_info)
* to be released by the balloon driver.
*/
if (trylock_page(page)) {
+#ifdef CONFIG_BALLOON_COMPACTION
if (!PagePrivate(page)) {
/* raced with isolation */
unlock_page(page);
continue;
}
+#endif
spin_lock_irqsave(&b_dev_info->pages_lock, flags);
balloon_page_delete(page);
__count_vm_event(BALLOON_DEFLATE);
On Wed, Oct 29, 2014 at 02:51:07PM +0400, Konstantin Khlebnikov wrote:
> Fix for commit d6d86c0a7f8ddc5b38cf089222cb1d9540762dc2
> ("mm/balloon_compaction: redesign ballooned pages management").
>
> If CONFIG_BALLOON_COMPACTION=n balloon_page_insert() does not link
> pages with balloon and doesn't set PagePrivate flag, as a result
> balloon_page_dequeue cannot get any pages because it thinks that
> all of them are isolated. Without balloon compaction nobody can
> isolate ballooned pages, it's safe to remove this check.
>
> Signed-off-by: Konstantin Khlebnikov <[email protected]>
> Reported-by: Matt Mullins <[email protected]>
> Cc: Stable <[email protected]> (v3.17)
That seems to do it, thanks!
Tested-by: Matt Mullins <[email protected]>