Re: [PATCH v3 00/20] vmw_balloon: compaction, shrinker, 64-bit, etc.

[gregkh@xxxxxxxxxxxxxxxxxxx]

On Tue, Oct 30, 2018 at 04:32:22PM +0000, Nadav Amit wrote:
> From: Nadav Amit
> Sent: September 26, 2018 at 7:13:16 PM GMT
> > To: Arnd Bergmann <arnd@xxxxxxxx>, gregkh@xxxxxxxxxxxxxxxxxxx
> > Cc: Xavier Deguillard <xdeguillard@xxxxxxxxxx>, linux-kernel@xxxxxxxxxxxxxxx>, Nadav Amit <namit@xxxxxxxxxx>, Michael S. Tsirkin <mst@xxxxxxxxxx>, Jason Wang <jasowang@xxxxxxxxxx>, linux-mm@xxxxxxxxx>, virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx
> > Subject: [PATCH v3 00/20] vmw_balloon: compaction, shrinker, 64-bit, etc.
> > 
> > 
> > This patch-set adds the following enhancements to the VMware balloon
> > driver:
> > 
> > 1. Balloon compaction support.
> > 2. Report the number of inflated/deflated ballooned pages through vmstat.
> > 3. Memory shrinker to avoid balloon over-inflation (and OOM).
> > 4. Support VMs with memory limit that is greater than 16TB.
> > 5. Faster and more aggressive inflation.
> > 
> > To support compaction we wish to use the existing infrastructure.
> > However, we need to make slight adaptions for it. We add a new list
> > interface to balloon-compaction, which is more generic and efficient,
> > since it does not require as many IRQ save/restore operations. We leave
> > the old interface that is used by the virtio balloon.
> > 
> > Big parts of this patch-set are cleanup and documentation. Patches 1-13
> > simplify the balloon code, document its behavior and allow the balloon
> > code to run concurrently. The support for concurrency is required for
> > compaction and the shrinker interface.
> > 
> > For documentation we use the kernel-doc format. We are aware that the
> > balloon interface is not public, but following the kernel-doc format may
> > be useful one day.
> > 
> > v2->v3: * Moving the balloon magic-number out of uapi (Greg)
> > 
> > v1->v2:	* Fix build error when THP is off (kbuild)
> > 	* Fix build error on i386 (kbuild)
> > 
> 
> Greg,
> 
> I realize you didnât apply patches 17-20. Any reason for that?

I have no idea, that was a few thousand patches reviewed ago...

Did I not say anything about this when I applied them?

greg k-h