On Tue, 2015-06-23 at 07:32 +0200, Michael S. Tsirkin wrote: > On Mon, Jun 22, 2015 at 02:09:48PM -0700, James Bottomley wrote: > > On Mon, 2015-06-22 at 15:56 +0300, Denis V. Lunev wrote: > > > On 15/06/15 13:52, Denis V. Lunev wrote: > > > > Excessive virtio_balloon inflation can cause invocation of OOM-killer, > > > > when Linux is under severe memory pressure. Various mechanisms are > > > > responsible for correct virtio_balloon memory management. Nevertheless > > > > it > > > > is often the case that these control tools does not have enough time to > > > > react on fast changing memory load. As a result OS runs out of memory > > > > and > > > > invokes OOM-killer. The balancing of memory by use of the virtio balloon > > > > should not cause the termination of processes while there are pages in > > > > the > > > > balloon. Now there is no way for virtio balloon driver to free memory at > > > > the last moment before some process get killed by OOM-killer. > > > > > > > > This does not provide a security breach as balloon itself is running > > > > inside Guest OS and is working in the cooperation with the host. Thus > > > > some improvements from Guest side should be considered as normal. > > > > > > > > To solve the problem, introduce a virtio_balloon callback which is > > > > expected to be called from the oom notifier call chain in > > > > out_of_memory() > > > > function. If virtio balloon could release some memory, it will make the > > > > system to return and retry the allocation that forced the out of memory > > > > killer to run. > > > > > > > > This behavior should be enabled if and only if appropriate feature bit > > > > is set on the device. It is off by default. > > > > > > > > This functionality was recently merged into vanilla Linux (actually in > > > > linux-next at the moment) > > > > > > > > commit 5a10b7dbf904bfe01bb9fcc6298f7df09eed77d5 > > > > Author: Raushaniya Maksudova <rmaksud...@parallels.com> > > > > Date: Mon Nov 10 09:36:29 2014 +1030 > > > > > > > > This patch adds respective control bits into QEMU. It introduces > > > > deflate-on-oom option for baloon device which do the trick. > > > > > > > > Changes from v6: > > > > - ported to virtio_add_feature > > > > > > > > Changes from v5: > > > > - ported to QEMU current > > > > > > > > Changes from v4: > > > > - spelling corrected according to suggestions from Eric Blake > > > > > > > > Changes from v3: > > > > - ported to git://git.kernel.org/pub/scm/virt/kvm/mst/qemu.git > > > > tags/for_upstream_rebased > > > > > > > > Changes from v2: > > > > - fixed mistake with bit number in virtio_balloon_get_features > > > > > > > > Changes from v1: > > > > - From: in patch 1 according to the original ownership > > > > - feature processing in patch 2 as suggested by Michael. It could be > > > > done > > > > without additional field, but this will require to move the property > > > > level up, i.e. to PCI & CCW level. > > > > > > > > Signed-off-by: Raushaniya Maksudova <rmaksud...@virtuozzo.com> > > > > Signed-off-by: Denis V. Lunev <d...@openvz.org> > > > > CC: Anthony Liguori <aligu...@amazon.com> > > > > CC: Michael S. Tsirkin <m...@redhat.com> > > > > > > > > P.S. Sorry for resend (if you have caught additional patch), I have > > > > expirienced some troubles in the process > > > > > > > ping > > > > We seem to have become bogged down in a dispute over whether this flag > > should be automatically enabled or disabled. To be honest, we don't > > care, since we're going to be shipping qemu configured according to our > > requirements (as are most other distros anyway). > > > > However, for the sake of getting the patch in, what about putting it in > > as is (default disable) because that has no impact to the status quo. > > If it later turns out everyone ships in a default enabled configuration, > > why then someone can patch upstream qemu to match. > > > > James > > > > Great, so I get your ack on it?
Sure, but I wasn't aware you were waiting for it Acked-by: James Bottomley <jbottom...@odin.com> James
