From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.9 required=3.0 tests=DKIMWL_WL_HIGH,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH, MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id BDAE3C2BB1D for ; Wed, 11 Mar 2020 20:23:08 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id 9DF542074A for ; Wed, 11 Mar 2020 20:23:09 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (1024-bit key) header.d=redhat.com header.i=@redhat.com header.b="IKGZksAT" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 9DF542074A Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=redhat.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=owner-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix) id CB9DF6B0005; Wed, 11 Mar 2020 16:23:07 -0400 (EDT) Received: by kanga.kvack.org (Postfix, from userid 40) id C68CB6B0006; Wed, 11 Mar 2020 16:23:07 -0400 (EDT) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id AE4106B0007; Wed, 11 Mar 2020 16:23:07 -0400 (EDT) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0082.hostedemail.com [216.40.44.82]) by kanga.kvack.org (Postfix) with ESMTP id 904796B0005 for ; Wed, 11 Mar 2020 16:23:07 -0400 (EDT) Received: from smtpin06.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay01.hostedemail.com (Postfix) with ESMTP id 4B081180AC497 for ; Wed, 11 Mar 2020 20:23:07 +0000 (UTC) X-FDA: 76584205614.06.night13_1b2fbb244c228 X-HE-Tag: night13_1b2fbb244c228 X-Filterd-Recvd-Size: 16441 Received: from us-smtp-delivery-1.mimecast.com (us-smtp-2.mimecast.com [207.211.31.81]) by imf16.hostedemail.com (Postfix) with ESMTP for ; Wed, 11 Mar 2020 20:23:06 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1583958186; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=6dOiOI2DppyNJLn/fltPLn4+IzBS890JpIZCzERAEiE=; b=IKGZksATsddJjreHCuZS4VoFPcch7Fi4FPf4y+t6dM1lIorgD4onhtM4U2F89gx8VcEXdJ Q+wLn6lTxQa280rJO7RPCwPmB8LggO7DNIlnk0x0sE6zQIEAX76FkVqOXz7wt6+FRjUk00 gLDWnrYMctKgjz1ORNa7HTZ2u251N10= Received: from mail-qk1-f197.google.com (mail-qk1-f197.google.com [209.85.222.197]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-179-_mBrEYlpOrqGHlJ5hXfxdA-1; Wed, 11 Mar 2020 16:22:59 -0400 X-MC-Unique: _mBrEYlpOrqGHlJ5hXfxdA-1 Received: by mail-qk1-f197.google.com with SMTP id d14so2313984qkj.5 for ; Wed, 11 Mar 2020 13:22:59 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:from:to:cc:subject:message-id:references :mime-version:content-disposition:in-reply-to; bh=PdHceSImpOs6CVFH+4WWRz5XkKao35jOzXo+8pbbF30=; b=TcBrr9nQAkROnQDjOl8YYdVLeXjj06+Guzv8Q3jFFlyw29FeMRyBwCUcC742sJ0fkG XVD375gBbbyXn/YHebZIJT7GyMh6V0OckzWsHNXTKuYWdykV5iv7X2AN3H6qiqKfWufO sBZik05mFKAQh6PUi1nPBDle9mfYo8gKb9gxd/ajTo6TCciTdQydPoRve+R28XNbYZbv teIHfQCCM8B1bHidMkVxr+O04BaHpj50CzUN1As0s8VB6NSbWzb9CM/0YyjkBH4Epyff T+yoqVvrAcL3spzPeqrcpUIu6U++Qp77JqfG8U7KfSwRbpbyfm+31NUvwgheNcbWzbBS 5RSA== X-Gm-Message-State: ANhLgQ1uwDkzbBBdinJnl85qfDYWpN4ZXyhzJGvI68l0tjwUEmMH+eb+ OIkR0YZJfkOtDy6q8fvehbuFoFDSSTxjavO0UGdNyyBz0EsUCG21NVoH1+pOl/zdRwHbKwrjq7s ARSRKtwZv0v4= X-Received: by 2002:ac8:2bf9:: with SMTP id n54mr4341669qtn.280.1583958179156; Wed, 11 Mar 2020 13:22:59 -0700 (PDT) X-Google-Smtp-Source: ADFU+vtYMiFrpDX7axDp/0tN80VnUfNwhdwRFbVcBrpyR+sVm+DN74jHBXdH09jbMEinw1MRgIwyBQ== X-Received: by 2002:ac8:2bf9:: with SMTP id n54mr4341645qtn.280.1583958178798; Wed, 11 Mar 2020 13:22:58 -0700 (PDT) Received: from redhat.com (bzq-79-178-2-19.red.bezeqint.net. [79.178.2.19]) by smtp.gmail.com with ESMTPSA id x6sm6089374qke.43.2020.03.11.13.22.54 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 11 Mar 2020 13:22:58 -0700 (PDT) Date: Wed, 11 Mar 2020 16:22:51 -0400 From: "Michael S. Tsirkin" To: David Hildenbrand Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org, virtualization@lists.linux-foundation.org, Jason Wang , Stephen Rothwell , Tyler Sanderson , David Rientjes , Wei Wang , Alexander Duyck , Nadav Amit , Michal Hocko , Andrew Morton Subject: Re: [PATCH v4] virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM Message-ID: <20200311162218-mutt-send-email-mst@kernel.org> References: <20200311135523.18512-1-david@redhat.com> <20200311135523.18512-2-david@redhat.com> MIME-Version: 1.0 In-Reply-To: <20200311135523.18512-2-david@redhat.com> X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable Content-Disposition: inline X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: On Wed, Mar 11, 2020 at 02:55:23PM +0100, 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. >=20 > However, the balloon is not simply some other slab cache that should be > shrunk when under memory pressure. The shrinker does not have a concept o= f > priorities yet, so this behavior cannot be configured. Eventually once > that is in place, we might want to switch back after doing proper > testing. >=20 > There was a report that this results in undesired side effects when > inflating the balloon to shrink the page cache. [1] > =09"When inflating the balloon against page cache (i.e. no free memory > =09 remains) vmscan.c will both shrink page cache, but also invoke the > =09 shrinkers -- including the balloon's shrinker. So the balloon > =09 driver allocates memory which requires reclaim, vmscan gets this > =09 memory by shrinking the balloon, and then the driver adds the > =09 memory back to the balloon. Basically a busy no-op." >=20 > 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. >=20 > Keep using the shrinker for VIRTIO_BALLOON_F_FREE_PAGE_HINT, because > this has no such side effects. Always register the shrinker with > VIRTIO_BALLOON_F_FREE_PAGE_HINT now. We are always allowed to reuse free > pages that are still to be processed by the guest. The hypervisor takes > care of identifying and resolving possible races between processing a > hinting request and the guest reusing a page. >=20 > In contrast to pre commit 71994620bb25 ("virtio_balloon: replace oom > notifier with shrinker"), don't add a module parameter to configure the > number of pages to deflate on OOM. Can be re-added if really needed. > Also, pay attention that leak_balloon() returns the number of 4k pages - > convert it properly in virtio_balloon_oom_notify(). >=20 > Testing done by Tyler for future reference: > Test setup: VM with 16 CPU, 64GB RAM. Running Debian 10. We have a 42 > GB file full of random bytes that we continually cat to /dev/null. > This fills the page cache as the file is read. Meanwhile, we trigger > the balloon to inflate, with a target size of 53 GB. This setup causes > the balloon inflation to pressure the page cache as the page cache is > also trying to grow. Afterwards we shrink the balloon back to zero (so > total deflate =3D total inflate). >=20 > Without this patch (kernel 4.19.0-5): > Inflation never reaches the target until we stop the "cat file > > /dev/null" process. Total inflation time was 542 seconds. The longest > period that made no net forward progress was 315 seconds. > Result of "grep balloon /proc/vmstat" after the test: > balloon_inflate 154828377 > balloon_deflate 154828377 >=20 > With this patch (kernel 5.6.0-rc4+): > Total inflation duration was 63 seconds. No deflate-queue activity > occurs when pressuring the page-cache. > Result of "grep balloon /proc/vmstat" after the test: > balloon_inflate 12968539 > balloon_deflate 12968539 >=20 > Conclusion: This patch fixes the issue. In the test it reduced > inflate/deflate activity by 12x, and reduced inflation time by 8.6x. > But more importantly, if we hadn't killed the "cat file > /dev/null" > process then, without the patch, the inflation process would never reac= h > the target. >=20 > [1] https://www.spinics.net/lists/linux-virtualization/msg40863.html >=20 > Reported-by: Tyler Sanderson > Tested-by: Tyler Sanderson > Acked-by: David Rientjes > Fixes: 71994620bb25 ("virtio_balloon: replace oom notifier with shrinker"= ) > Cc: Michael S. Tsirkin > Cc: Wei Wang > Cc: Alexander Duyck > Cc: David Rientjes > Cc: Nadav Amit > Cc: Michal Hocko > Cc: Andrew Morton > Signed-off-by: David Hildenbrand Acked-by: Michael S. Tsirkin Andrew, pls merge through your tree. > --- > drivers/virtio/virtio_balloon.c | 103 +++++++++++++++----------------- > 1 file changed, 47 insertions(+), 56 deletions(-) >=20 > diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_ball= oon.c > index 9e0177529bad..0ef16566c3f3 100644 > --- a/drivers/virtio/virtio_balloon.c > +++ b/drivers/virtio/virtio_balloon.c > @@ -14,6 +14,7 @@ > #include > #include > #include > +#include > #include > #include > #include > @@ -28,7 +29,9 @@ > */ > #define VIRTIO_BALLOON_PAGES_PER_PAGE (unsigned)(PAGE_SIZE >> VIRTIO_BAL= LOON_PFN_SHIFT) > #define VIRTIO_BALLOON_ARRAY_PFNS_MAX 256 > -#define VIRTBALLOON_OOM_NOTIFY_PRIORITY 80 > +/* Maximum number of (4k) pages to deflate on OOM notifications. */ > +#define VIRTIO_BALLOON_OOM_NR_PAGES 256 > +#define VIRTIO_BALLOON_OOM_NOTIFY_PRIORITY 80 > =20 > #define VIRTIO_BALLOON_FREE_PAGE_ALLOC_FLAG (__GFP_NORETRY | __GFP_NOWAR= N | \ > =09=09=09=09=09 __GFP_NOMEMALLOC) > @@ -114,9 +117,12 @@ struct virtio_balloon { > =09/* Memory statistics */ > =09struct virtio_balloon_stat stats[VIRTIO_BALLOON_S_NR]; > =20 > -=09/* To register a shrinker to shrink memory upon memory pressure */ > +=09/* Shrinker to return free pages - VIRTIO_BALLOON_F_FREE_PAGE_HINT */ > =09struct shrinker shrinker; > =20 > +=09/* OOM notifier to deflate on OOM - VIRTIO_BALLOON_F_DEFLATE_ON_OOM *= / > +=09struct notifier_block oom_nb; > + > =09/* Free page reporting device */ > =09struct virtqueue *reporting_vq; > =09struct page_reporting_dev_info pr_dev_info; > @@ -830,50 +836,13 @@ static unsigned long shrink_free_pages(struct virti= o_balloon *vb, > =09return blocks_freed * VIRTIO_BALLOON_HINT_BLOCK_PAGES; > } > =20 > -static unsigned long leak_balloon_pages(struct virtio_balloon *vb, > - unsigned long pages_to_free) > -{ > -=09return leak_balloon(vb, pages_to_free * VIRTIO_BALLOON_PAGES_PER_PAGE= ) / > -=09=09VIRTIO_BALLOON_PAGES_PER_PAGE; > -} > - > -static unsigned long shrink_balloon_pages(struct virtio_balloon *vb, > -=09=09=09=09=09 unsigned long pages_to_free) > -{ > -=09unsigned long pages_freed =3D 0; > - > -=09/* > -=09 * One invocation of leak_balloon can deflate at most > -=09 * VIRTIO_BALLOON_ARRAY_PFNS_MAX balloon pages, so we call it > -=09 * multiple times to deflate pages till reaching pages_to_free. > -=09 */ > -=09while (vb->num_pages && pages_freed < pages_to_free) > -=09=09pages_freed +=3D leak_balloon_pages(vb, > -=09=09=09=09=09=09 pages_to_free - pages_freed); > - > -=09update_balloon_size(vb); > - > -=09return pages_freed; > -} > - > static unsigned long virtio_balloon_shrinker_scan(struct shrinker *shrin= ker, > =09=09=09=09=09=09 struct shrink_control *sc) > { > -=09unsigned long pages_to_free, pages_freed =3D 0; > =09struct virtio_balloon *vb =3D container_of(shrinker, > =09=09=09=09=09struct virtio_balloon, shrinker); > =20 > -=09pages_to_free =3D sc->nr_to_scan; > - > -=09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_FREE_PAGE_HINT)) > -=09=09pages_freed =3D shrink_free_pages(vb, pages_to_free); > - > -=09if (pages_freed >=3D pages_to_free) > -=09=09return pages_freed; > - > -=09pages_freed +=3D shrink_balloon_pages(vb, pages_to_free - pages_freed= ); > - > -=09return pages_freed; > +=09return shrink_free_pages(vb, sc->nr_to_scan); > } > =20 > static unsigned long virtio_balloon_shrinker_count(struct shrinker *shri= nker, > @@ -881,12 +850,22 @@ static unsigned long virtio_balloon_shrinker_count(= struct shrinker *shrinker, > { > =09struct virtio_balloon *vb =3D container_of(shrinker, > =09=09=09=09=09struct virtio_balloon, shrinker); > -=09unsigned long count; > =20 > -=09count =3D vb->num_pages / VIRTIO_BALLOON_PAGES_PER_PAGE; > -=09count +=3D vb->num_free_page_blocks * VIRTIO_BALLOON_HINT_BLOCK_PAGES= ; > +=09return vb->num_free_page_blocks * VIRTIO_BALLOON_HINT_BLOCK_PAGES; > +} > + > +static int virtio_balloon_oom_notify(struct notifier_block *nb, > +=09=09=09=09 unsigned long dummy, void *parm) > +{ > +=09struct virtio_balloon *vb =3D container_of(nb, > +=09=09=09=09=09=09 struct virtio_balloon, oom_nb); > +=09unsigned long *freed =3D parm; > + > +=09*freed +=3D leak_balloon(vb, VIRTIO_BALLOON_OOM_NR_PAGES) / > +=09=09 VIRTIO_BALLOON_PAGES_PER_PAGE; > +=09update_balloon_size(vb); > =20 > -=09return count; > +=09return NOTIFY_OK; > } > =20 > static void virtio_balloon_unregister_shrinker(struct virtio_balloon *vb= ) > @@ -971,7 +950,23 @@ static int virtballoon_probe(struct virtio_device *v= dev) > =09=09=09=09=09=09 VIRTIO_BALLOON_CMD_ID_STOP); > =09=09spin_lock_init(&vb->free_page_list_lock); > =09=09INIT_LIST_HEAD(&vb->free_page_list); > +=09=09/* > +=09=09 * We're allowed to reuse any free pages, even if they are > +=09=09 * still to be processed by the host. > +=09=09 */ > +=09=09err =3D virtio_balloon_register_shrinker(vb); > +=09=09if (err) > +=09=09=09goto out_del_balloon_wq; > +=09} > + > +=09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_DEFLATE_ON_OOM)) { > +=09=09vb->oom_nb.notifier_call =3D virtio_balloon_oom_notify; > +=09=09vb->oom_nb.priority =3D VIRTIO_BALLOON_OOM_NOTIFY_PRIORITY; > +=09=09err =3D register_oom_notifier(&vb->oom_nb); > +=09=09if (err < 0) > +=09=09=09goto out_unregister_shrinker; > =09} > + > =09if (virtio_has_feature(vdev, VIRTIO_BALLOON_F_PAGE_POISON)) { > =09=09/* Start with poison val of 0 representing general init */ > =09=09__u32 poison_val =3D 0; > @@ -986,15 +981,6 @@ static int virtballoon_probe(struct virtio_device *v= dev) > =09=09virtio_cwrite(vb->vdev, struct virtio_balloon_config, > =09=09=09 poison_val, &poison_val); > =09} > -=09/* > -=09 * We continue to use VIRTIO_BALLOON_F_DEFLATE_ON_OOM to decide if a > -=09 * shrinker needs to be registered to relieve memory pressure. > -=09 */ > -=09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_DEFLATE_ON_OOM)) { > -=09=09err =3D virtio_balloon_register_shrinker(vb); > -=09=09if (err) > -=09=09=09goto out_del_balloon_wq; > -=09} > =20 > =09vb->pr_dev_info.report =3D virtballoon_free_page_report; > =09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_REPORTING)) { > @@ -1003,12 +989,12 @@ static int virtballoon_probe(struct virtio_device = *vdev) > =09=09capacity =3D virtqueue_get_vring_size(vb->reporting_vq); > =09=09if (capacity < PAGE_REPORTING_CAPACITY) { > =09=09=09err =3D -ENOSPC; > -=09=09=09goto out_unregister_shrinker; > +=09=09=09goto out_unregister_oom; > =09=09} > =20 > =09=09err =3D page_reporting_register(&vb->pr_dev_info); > =09=09if (err) > -=09=09=09goto out_unregister_shrinker; > +=09=09=09goto out_unregister_oom; > =09} > =20 > =09virtio_device_ready(vdev); > @@ -1017,8 +1003,11 @@ static int virtballoon_probe(struct virtio_device = *vdev) > =09=09virtballoon_changed(vdev); > =09return 0; > =20 > -out_unregister_shrinker: > +out_unregister_oom: > =09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_DEFLATE_ON_OOM)) > +=09=09unregister_oom_notifier(&vb->oom_nb); > +out_unregister_shrinker: > +=09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_FREE_PAGE_HINT)) > =09=09virtio_balloon_unregister_shrinker(vb); > out_del_balloon_wq: > =09if (virtio_has_feature(vdev, VIRTIO_BALLOON_F_FREE_PAGE_HINT)) > @@ -1061,6 +1050,8 @@ static void virtballoon_remove(struct virtio_device= *vdev) > =09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_REPORTING)) > =09=09page_reporting_unregister(&vb->pr_dev_info); > =09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_DEFLATE_ON_OOM)) > +=09=09unregister_oom_notifier(&vb->oom_nb); > +=09if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_FREE_PAGE_HINT)) > =09=09virtio_balloon_unregister_shrinker(vb); > =09spin_lock_irq(&vb->stop_update_lock); > =09vb->stop_update =3D true; > --=20 > 2.24.1