Re: [Xen-devel] [PATCH v2 2/2] xen/balloon: Enforce various limits on target

[Ian Campbell <Ian.Campbell@citrix.com>]

On Tue, 2013-04-30 at 19:58 +0100, Daniel Kiper wrote:
> On Tue, Apr 30, 2013 at 02:44:18PM +0100, Ian Campbell wrote:
> > On Tue, 2013-04-30 at 13:59 +0100, Daniel Kiper wrote:
> > > On Mon, Apr 29, 2013 at 03:44:09PM +0100, Ian Campbell wrote:
> > > > On Mon, 2013-04-29 at 12:37 +0100, Daniel Kiper wrote:
> > > >
> > > > > This patch enforces on target limit statically defined in Linux Kernel
> > > > > source and limit defined by hypervisor or host. This way the balloon
> > > > > driver should not attempt to populate pages above given limits
> > > > > because they may fail.
> > > > >
> > > > > Particularly this patch fixes bug which led to flood
> > > > > of dom0 kernel log with messages similar to:
> > > > >
> > > > > System RAM resource [mem 0x1b8000000-0x1bfffffff] cannot be added
> > > > > xen_balloon: reserve_additional_memory: add_memory() failed: -17
> > > >
> > > > I think it would be OK to simply tone down this message (and perhaps add
> > > > the failed pages to the balloon, if that makes sense). This isn't
> > > > dissimilar to increase_reservation failing.
> > >
> > > If add_memory() fails it is hard error. It means that we do not
> > > know where new or ballooned pages should be placed.
> >
> > I see that add_memory() is a generic or arch level function rather than
> > a ballooning specific one. Under what circumstances can it fail and how
> > do they relate the the setting of the balloon target?
> 
> It is generic function with some references to arch code. It is called
> when pages could not be taken from balloon any more and must be hotplugged.
> It reserves memory resource (start address and size is calculated on the
> base of target and max_pfn) and build some structures for new memory.
> It may fail in many places. In this case it failed because placement
> of new resource was incorrectly calculated because algorithm is very
> simple and not cover all cases. I am going to fix this issue but
> a bit later.

I think you should fix that core issue first, otherwise the requirement
for messing with the balloon target is not very obvious.

Ian.