From: Michal Hocko <mhocko@kernel.org> To: David Hildenbrand <david@redhat.com> Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org, virtualization@lists.linux-foundation.org, Tyler Sanderson <tysand@google.com>, "Michael S . Tsirkin" <mst@redhat.com>, Wei Wang <wei.w.wang@intel.com>, Alexander Duyck <alexander.h.duyck@linux.intel.com>, David Rientjes <rientjes@google.com>, Nadav Amit <namit@vmware.com> Subject: Re: [PATCH v1 3/3] virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM Date: Fri, 14 Feb 2020 15:06:41 +0100 [thread overview] Message-ID: <20200214140641.GB31689@dhcp22.suse.cz> (raw) In-Reply-To: <20200205163402.42627-4-david@redhat.com> On Wed 05-02-20 17:34:02, David Hildenbrand wrote: > Commit 71994620bb25 ("virtio_balloon: replace oom notifier with shrinker") > changed the behavior when deflation happens automatically. Instead of > deflating when called by the OOM handler, the shrinker is used. > > However, the balloon is not simply some slab cache that should be > shrunk when under memory pressure. The shrinker does not have a concept of > priorities, so this behavior cannot be configured. Adding a priority to the shrinker doesn't sound like a big problem to me. Shrinkers already get shrink_control data structure already and priority could be added there. > There was a report that this results in undesired side effects when > inflating the balloon to shrink the page cache. [1] > "When inflating the balloon against page cache (i.e. no free memory > remains) vmscan.c will both shrink page cache, but also invoke the > shrinkers -- including the balloon's shrinker. So the balloon > driver allocates memory which requires reclaim, vmscan gets this > memory by shrinking the balloon, and then the driver adds the > memory back to the balloon. Basically a busy no-op." > > The name "deflate on OOM" makes it pretty clear when deflation should > happen - after other approaches to reclaim memory failed, not while > reclaiming. This allows to minimize the footprint of a guest - memory > will only be taken out of the balloon when really needed. > > Especially, a drop_slab() will result in the whole balloon getting > deflated - undesired. Could you explain why some more? drop_caches shouldn't be really used in any production workloads and if somebody really wants all the cache to be dropped then why is balloon any different? -- Michal Hocko SUSE Labs
WARNING: multiple messages have this Message-ID (diff)
From: Michal Hocko <mhocko@kernel.org> To: David Hildenbrand <david@redhat.com> Cc: "Michael S . Tsirkin" <mst@redhat.com>, linux-kernel@vger.kernel.org, virtualization@lists.linux-foundation.org, Tyler Sanderson <tysand@google.com>, linux-mm@kvack.org, Nadav Amit <namit@vmware.com>, David Rientjes <rientjes@google.com>, Alexander Duyck <alexander.h.duyck@linux.intel.com> Subject: Re: [PATCH v1 3/3] virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM Date: Fri, 14 Feb 2020 15:06:41 +0100 [thread overview] Message-ID: <20200214140641.GB31689@dhcp22.suse.cz> (raw) In-Reply-To: <20200205163402.42627-4-david@redhat.com> On Wed 05-02-20 17:34:02, David Hildenbrand wrote: > Commit 71994620bb25 ("virtio_balloon: replace oom notifier with shrinker") > changed the behavior when deflation happens automatically. Instead of > deflating when called by the OOM handler, the shrinker is used. > > However, the balloon is not simply some slab cache that should be > shrunk when under memory pressure. The shrinker does not have a concept of > priorities, so this behavior cannot be configured. Adding a priority to the shrinker doesn't sound like a big problem to me. Shrinkers already get shrink_control data structure already and priority could be added there. > There was a report that this results in undesired side effects when > inflating the balloon to shrink the page cache. [1] > "When inflating the balloon against page cache (i.e. no free memory > remains) vmscan.c will both shrink page cache, but also invoke the > shrinkers -- including the balloon's shrinker. So the balloon > driver allocates memory which requires reclaim, vmscan gets this > memory by shrinking the balloon, and then the driver adds the > memory back to the balloon. Basically a busy no-op." > > The name "deflate on OOM" makes it pretty clear when deflation should > happen - after other approaches to reclaim memory failed, not while > reclaiming. This allows to minimize the footprint of a guest - memory > will only be taken out of the balloon when really needed. > > Especially, a drop_slab() will result in the whole balloon getting > deflated - undesired. Could you explain why some more? drop_caches shouldn't be really used in any production workloads and if somebody really wants all the cache to be dropped then why is balloon any different? -- Michal Hocko SUSE Labs
next prev parent reply other threads:[~2020-02-14 14:06 UTC|newest] Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-02-05 16:33 [PATCH v1 0/3] virtio-balloon: Fixes + switch back to OOM handler David Hildenbrand 2020-02-05 16:33 ` David Hildenbrand 2020-02-05 16:34 ` [PATCH v1 1/3] virtio-balloon: Fix memory leak when unloading while hinting is in progress David Hildenbrand 2020-02-06 8:36 ` Michael S. Tsirkin 2020-02-05 16:34 ` [PATCH v1 2/3] virtio_balloon: Fix memory leaks on errors in virtballoon_probe() David Hildenbrand 2020-02-06 8:36 ` Michael S. Tsirkin 2020-02-05 16:34 ` [PATCH v1 3/3] virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM David Hildenbrand 2020-02-05 22:37 ` Tyler Sanderson 2020-02-05 22:37 ` Tyler Sanderson via Virtualization 2020-02-05 22:52 ` David Hildenbrand 2020-02-05 22:52 ` David Hildenbrand 2020-02-05 23:06 ` Tyler Sanderson 2020-02-05 23:06 ` Tyler Sanderson via Virtualization 2020-02-06 7:40 ` Michael S. Tsirkin 2020-02-06 7:40 ` Michael S. Tsirkin 2020-02-06 8:42 ` David Hildenbrand 2020-02-06 8:57 ` Michael S. Tsirkin 2020-02-06 9:05 ` David Hildenbrand 2020-02-06 9:09 ` Michael S. Tsirkin 2020-02-06 8:57 ` Wang, Wei W 2020-02-06 8:57 ` Wang, Wei W 2020-02-06 9:11 ` Michael S. Tsirkin 2020-02-06 9:12 ` Michael S. Tsirkin 2020-02-06 9:21 ` David Hildenbrand 2020-02-14 9:51 ` David Hildenbrand 2020-02-14 13:31 ` Wang, Wei W 2020-02-14 13:31 ` Wang, Wei W 2020-02-16 9:47 ` Michael S. Tsirkin 2020-02-16 9:47 ` Michael S. Tsirkin 2020-02-21 3:29 ` Tyler Sanderson 2020-02-21 3:29 ` Tyler Sanderson via Virtualization 2020-03-08 4:47 ` Tyler Sanderson 2020-03-08 4:47 ` Tyler Sanderson via Virtualization 2020-03-08 4:47 ` Tyler Sanderson 2020-03-09 9:03 ` David Hildenbrand 2020-03-09 10:14 ` Michael S. Tsirkin 2020-03-09 10:59 ` David Hildenbrand 2020-03-09 10:24 ` Michael S. Tsirkin 2020-02-14 14:06 ` Michal Hocko [this message] 2020-02-14 14:06 ` Michal Hocko 2020-02-14 14:18 ` David Hildenbrand 2020-02-14 20:48 ` Tyler Sanderson 2020-02-14 20:48 ` Tyler Sanderson via Virtualization 2020-02-14 21:17 ` David Hildenbrand 2020-02-14 21:17 ` David Hildenbrand 2020-02-16 9:46 ` Michael S. Tsirkin 2020-02-16 9:46 ` Michael S. Tsirkin
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20200214140641.GB31689@dhcp22.suse.cz \ --to=mhocko@kernel.org \ --cc=alexander.h.duyck@linux.intel.com \ --cc=david@redhat.com \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-mm@kvack.org \ --cc=mst@redhat.com \ --cc=namit@vmware.com \ --cc=rientjes@google.com \ --cc=tysand@google.com \ --cc=virtualization@lists.linux-foundation.org \ --cc=wei.w.wang@intel.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.