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=-6.8 required=3.0 tests=DKIMWL_WL_HIGH,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS,HTML_MESSAGE, 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 51D43C352A1 for ; Wed, 5 Feb 2020 22:52:16 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id E030024125 for ; Wed, 5 Feb 2020 22:52:15 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (1024-bit key) header.d=redhat.com header.i=@redhat.com header.b="Z2O8oneL" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org E030024125 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 781106B0003; Wed, 5 Feb 2020 17:52:15 -0500 (EST) Received: by kanga.kvack.org (Postfix, from userid 40) id 732476B0005; Wed, 5 Feb 2020 17:52:15 -0500 (EST) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 5D0EB6B0006; Wed, 5 Feb 2020 17:52:15 -0500 (EST) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0149.hostedemail.com [216.40.44.149]) by kanga.kvack.org (Postfix) with ESMTP id 3CE456B0003 for ; Wed, 5 Feb 2020 17:52:15 -0500 (EST) Received: from smtpin16.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay01.hostedemail.com (Postfix) with ESMTP id E83D2180AD802 for ; Wed, 5 Feb 2020 22:52:14 +0000 (UTC) X-FDA: 76457573388.16.straw77_138f7b2191c2f X-HE-Tag: straw77_138f7b2191c2f X-Filterd-Recvd-Size: 30028 Received: from us-smtp-delivery-1.mimecast.com (us-smtp-2.mimecast.com [205.139.110.61]) by imf08.hostedemail.com (Postfix) with ESMTP for ; Wed, 5 Feb 2020 22:52:14 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1580943132; 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: in-reply-to:in-reply-to:references:references; bh=hUs/DiidQTwioeVZwlx/n0fTyMj2i+Fl87Q2QU19NXE=; b=Z2O8oneLnATEhnWudgfD4yxTOTh3dPEh48dvnZDy9Nnqi98vuwxrrx5ZL+IUxm/sCWJx+F SPVpfaeVpYWfgmcgiIIV7QYt4A7u5qwssINxzDU+tgBM8XvCoc0rB8qmBDY73vmYaWcEv3 KFtZF7wIjGkwHTNJsuoTdXWVNauBQ08= Received: from mail-wm1-f71.google.com (mail-wm1-f71.google.com [209.85.128.71]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-64-ud13Mi42PLiu6-MbDnnUAQ-1; Wed, 05 Feb 2020 17:52:08 -0500 Received: by mail-wm1-f71.google.com with SMTP id l11so2555761wmi.0 for ; Wed, 05 Feb 2020 14:52:07 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:content-transfer-encoding:from:mime-version :subject:date:message-id:references:cc:in-reply-to:to; bh=wkjVhVa2NyTDmZ8/LFbZ5n3Su9BgrxDkJjKz3sXdxiw=; b=LfXYP7atUW32FSq6buPYyDlXaVKs03rg6RgrdpgHuAjD66cCiJOyaHd6yqjBNPV2am 2F8iV3hN02SEOn4WqdcrSkaL64CV5o4HTxMWep9hm6ika4V57AWuJWqUV/r33/g0P0/V mPziA6Zh5s6Yid0g7DD6yuYDSPqj2m/ueQqme7kdFDoa84OBtZv0tU+xxVL4gFg2lZYr VGVeTp1qR6EhIwszfP9xje8BCK7QBBnFHLgBKO5GlZhqz9C0MCJmVwUEfzsbxksx3N8G Z86thMkdCHr/t6guGCC/6SYsaF/+5J5U84rXXkx0xGXrpD4JZubhqgCBXzFURENyb+sv USbA== X-Gm-Message-State: APjAAAUpAdbTFeguNeH5l4C/0HG9/j0uEOLK9VLngeWnq2B4OfIoKzWC PDWfvEsoRecwYotkj1Ev5jyESnmuabTIudctMwXDfKtHo3b+i/rXGek6suFgCSgG0XrTz4fqhet 0R4G24+8eQyM= X-Received: by 2002:a05:600c:217:: with SMTP id 23mr8022430wmi.124.1580943126911; Wed, 05 Feb 2020 14:52:06 -0800 (PST) X-Google-Smtp-Source: APXvYqwiDHT0qXGq2gA8EXYwjjj/RqXgsYsUlzIIAtdGXpOlDxU+MsMWas/ug+dwaury9C1t3uoRBA== X-Received: by 2002:a05:600c:217:: with SMTP id 23mr8022405wmi.124.1580943126587; Wed, 05 Feb 2020 14:52:06 -0800 (PST) Received: from [192.168.3.122] (p5B0C6AA4.dip0.t-ipconnect.de. [91.12.106.164]) by smtp.gmail.com with ESMTPSA id z19sm1144722wmi.35.2020.02.05.14.52.05 (version=TLS1_3 cipher=TLS_AES_128_GCM_SHA256 bits=128/128); Wed, 05 Feb 2020 14:52:06 -0800 (PST) From: David Hildenbrand Mime-Version: 1.0 (1.0) Subject: Re: [PATCH v1 3/3] virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM Date: Wed, 5 Feb 2020 23:52:05 +0100 Message-Id: <7C577D41-AC5C-4094-A280-CBA6ED05C9FB@redhat.com> References: Cc: David Hildenbrand , linux-kernel@vger.kernel.org, linux-mm@kvack.org, virtualization@lists.linux-foundation.org, "Michael S . Tsirkin" , Wei Wang , Alexander Duyck , David Rientjes , Nadav Amit , Michal Hocko In-Reply-To: To: Tyler Sanderson X-Mailer: iPhone Mail (17C54) X-MC-Unique: ud13Mi42PLiu6-MbDnnUAQ-1 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: multipart/alternative; boundary=Apple-Mail-F9A6D605-018C-4510-9EDF-244179F93224 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: --Apple-Mail-F9A6D605-018C-4510-9EDF-244179F93224 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable > Am 05.02.2020 um 23:37 schrieb Tyler Sanderson : >=20 > =EF=BB=BF >=20 >=20 >> On Wed, Feb 5, 2020 at 8:34 AM David Hildenbrand wrot= e: >> 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 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. >>=20 >> 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 mem= ory >> 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 thi= s >> memory by shrinking the balloon, and then the driver adds the >> 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 >> Especially, a drop_slab() will result in the whole balloon getting >> deflated - undesired. While handling it via the OOM handler might not be >> perfect, it keeps existing behavior. If we want a different behavior, th= en >> we need a new feature bit and document it properly (although, there shou= ld >> be a clear use case and the intended effects should be well described). >>=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 moodule 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 >> Note1: using the OOM handler is frowned upon, but it really is what we >> need for this feature. >>=20 >> Note2: without VIRTIO_BALLOON_F_MUST_TELL_HOST (iow, always with QEMU) w= e >> could actually skip sending deflation requests to our hypervisor, >> making the OOM path *very* simple. Besically freeing pages and >> updating the balloon. If the communication with the host ever >> becomes a problem on this call path. >>=20 >> [1] https://www.spinics.net/lists/linux-virtualization/msg40863.html >>=20 >> Reported-by: Tyler Sanderson >> Cc: Michael S. Tsirkin >> Cc: Wei Wang >> Cc: Alexander Duyck >> Cc: David Rientjes >> Cc: Nadav Amit >> Cc: Michal Hocko >> Signed-off-by: David Hildenbrand >> --- >> drivers/virtio/virtio_balloon.c | 107 +++++++++++++------------------- >> 1 file changed, 44 insertions(+), 63 deletions(-) >>=20 >> diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_bal= loon.c >> index 7e5d84caeb94..e7b18f556c5e 100644 >> --- a/drivers/virtio/virtio_balloon.c >> +++ b/drivers/virtio/virtio_balloon.c >> @@ -14,6 +14,7 @@ >> #include >> #include >> #include >> +#include >> #include >> #include >> #include >> @@ -27,7 +28,9 @@ >> */ >> #define VIRTIO_BALLOON_PAGES_PER_PAGE (unsigned)(PAGE_SIZE >> VIRTIO_BA= LLOON_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_NOWA= RN | \ >> __GFP_NOMEMALLOC) >> @@ -112,8 +115,11 @@ struct virtio_balloon { >> /* Memory statistics */ >> struct virtio_balloon_stat stats[VIRTIO_BALLOON_S_NR]; >>=20 >> - /* To register a shrinker to shrink memory upon memory pressure = */ >> + /* Shrinker to return free pages - VIRTIO_BALLOON_F_FREE_PAGE_HI= NT */ >> struct shrinker shrinker; >> + >> + /* OOM notifier to deflate on OOM - VIRTIO_BALLOON_F_DEFLATE_ON_= OOM */ >> + struct notifier_block oom_nb; >> }; >>=20 >> static struct virtio_device_id id_table[] =3D { >> @@ -786,50 +792,13 @@ static unsigned long shrink_free_pages(struct virt= io_balloon *vb, >> return blocks_freed * VIRTIO_BALLOON_HINT_BLOCK_PAGES; >> } >>=20 >> -static unsigned long leak_balloon_pages(struct virtio_balloon *vb, >> - unsigned long pages_to_free) >> -{ >> - return leak_balloon(vb, pages_to_free * VIRTIO_BALLOON_PAGES_PER= _PAGE) / >> - VIRTIO_BALLOON_PAGES_PER_PAGE; >> -} >> - >> -static unsigned long shrink_balloon_pages(struct virtio_balloon *vb, >> - unsigned long pages_to_free) >> -{ >> - unsigned long pages_freed =3D 0; >> - >> - /* >> - * One invocation of leak_balloon can deflate at most >> - * VIRTIO_BALLOON_ARRAY_PFNS_MAX balloon pages, so we call it >> - * multiple times to deflate pages till reaching pages_to_free. >> - */ >> - while (vb->num_pages && pages_freed < pages_to_free) >> - pages_freed +=3D leak_balloon_pages(vb, >> - pages_to_free - pages_= freed); >> - >> - update_balloon_size(vb); >> - >> - return pages_freed; >> -} >> - >> static unsigned long virtio_balloon_shrinker_scan(struct shrinker *shri= nker, >> struct shrink_control = *sc) >> { >> - unsigned long pages_to_free, pages_freed =3D 0; >> struct virtio_balloon *vb =3D container_of(shrinker, >> struct virtio_balloon, shrinker)= ; >>=20 >> - pages_to_free =3D sc->nr_to_scan; >> - >> - if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_FREE_PAGE_HINT= )) >> - pages_freed =3D shrink_free_pages(vb, pages_to_free); >> - >> - if (pages_freed >=3D pages_to_free) >> - return pages_freed; >> - >> - pages_freed +=3D shrink_balloon_pages(vb, pages_to_free - pages_= freed); >> - >> - return pages_freed; >> + return shrink_free_pages(vb, sc->nr_to_scan); >> } >>=20 >> static unsigned long virtio_balloon_shrinker_count(struct shrinker *shr= inker, >> @@ -837,26 +806,22 @@ static unsigned long virtio_balloon_shrinker_count= (struct shrinker *shrinker, >> { >> struct virtio_balloon *vb =3D container_of(shrinker, >> struct virtio_balloon, shrinker)= ; >> - unsigned long count; >> - >> - count =3D vb->num_pages / VIRTIO_BALLOON_PAGES_PER_PAGE; >> - count +=3D vb->num_free_page_blocks * VIRTIO_BALLOON_HINT_BLOCK_= PAGES; >>=20 >> - return count; >> + return vb->num_free_page_blocks * VIRTIO_BALLOON_HINT_BLOCK_PAGE= S; >> } >>=20 >> -static void virtio_balloon_unregister_shrinker(struct virtio_balloon *v= b) >> +static int virtio_balloon_oom_notify(struct notifier_block *nb, >> + unsigned long dummy, void *parm) >> { >> - unregister_shrinker(&vb->shrinker); >> -} >> + struct virtio_balloon *vb =3D container_of(nb, >> + struct virtio_balloon, = oom_nb); >> + unsigned long *freed =3D parm; >>=20 >> -static int virtio_balloon_register_shrinker(struct virtio_balloon *vb) >> -{ >> - vb->shrinker.scan_objects =3D virtio_balloon_shrinker_scan; >> - vb->shrinker.count_objects =3D virtio_balloon_shrinker_count; >> - vb->shrinker.seeks =3D DEFAULT_SEEKS; >> + *freed +=3D leak_balloon(vb, VIRTIO_BALLOON_OOM_NR_PAGES) / >> + VIRTIO_BALLOON_PAGES_PER_PAGE; >> + update_balloon_size(vb); >>=20 >> - return register_shrinker(&vb->shrinker); >> + return NOTIFY_OK; >> } >>=20 >> static int virtballoon_probe(struct virtio_device *vdev) >> @@ -933,22 +898,35 @@ static int virtballoon_probe(struct virtio_device = *vdev) >> virtio_cwrite(vb->vdev, struct virtio_balloon_co= nfig, >> poison_val, &poison_val); >> } >> - } >> - /* >> - * We continue to use VIRTIO_BALLOON_F_DEFLATE_ON_OOM to decide = if a >> - * shrinker needs to be registered to relieve memory pressure. >> - */ >> - if (virtio_has_feature(vb->vdev, VIRTIO_BALLOON_F_DEFLATE_ON_OOM= )) { >> - err =3D virtio_balloon_register_shrinker(vb); >> + >> + /* >> + * We're allowed to reuse any free pages, even if they a= re >> + * still to be processed by the host. > It is important to clarify that pages that are on the inflate queue but n= ot ACKed by the host (the queue entry has not been returned) are _not_ okay= to reuse. > If the host is going to do something destructive to the page (like deback= it) then that needs to happen before the entry is returned. While you are correct, this comment is in the =E2=80=9Efree page hinting=E2= =80=9C section/if (not obvious by looking at the diff only), so it does not= apply to inflate/deflate queues - but only free pages that are getting hin= ted. Or am I misreading your suggestion/missing something? Thanks! --Apple-Mail-F9A6D605-018C-4510-9EDF-244179F93224 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable


Am 05.02.2020 um 23:37 schrieb Tyler San= derson <tysand@google.com>:

=EF=BB=BF

=

= On Wed, Feb 5, 2020 at 8:34 AM David Hildenbrand <david@redhat.com> 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<= br> priorities, so this behavior cannot be configured.

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 c= ache, but also invoke the
         shrinkers -- including the balloon's shri= nker. So the balloon
         driver allocates memory which requires re= claim, vmscan gets this
         memory by shrinking the balloon, and then= the driver adds the
         memory back to the balloon. Basically a b= usy 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. While handling it via the OOM handler might not be perfect, it keeps existing behavior. If we want a different behavior, then<= br> we need a new feature bit and document it properly (although, there should<= br> be a clear use case and the intended effects should be well described).

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.

In contrast to pre commit 71994620bb25 ("virtio_balloon: replace oom
notifier with shrinker"), don't add a moodule 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().

Note1: using the OOM handler is frowned upon, but it really is what we
       need for this feature.

Note2: without VIRTIO_BALLOON_F_MUST_TELL_HOST (iow, always with QEMU) we        could actually skip sending deflation requests t= o our hypervisor,
       making the OOM path *very* simple. Besically fre= eing pages and
       updating the balloon. If the communication with = the host ever
       becomes a problem on this call path.

[1] https://www.spinics.net/lists/li= nux-virtualization/msg40863.html

Reported-by: Tyler Sanderson <tysand@google.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Wei Wang <= wei.w.wang@intel.com>
Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nadav Amit <na= mit@vmware.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_balloon.c | 107 +++++++++++++------------------= -
 1 file changed, 44 insertions(+), 63 deletions(-)

diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_balloo= n.c
index 7e5d84caeb94..e7b18f556c5e 100644
--- a/drivers/virtio/virtio_balloon.c
+++ b/drivers/virtio/virtio_balloon.c
@@ -14,6 +14,7 @@
 #include <linux/slab.h>
 #include <linux/module.h>
 #include <linux/balloon_compaction.h>
+#include <linux/oom.h>
 #include <linux/wait.h>
 #include <linux/mm.h>
 #include <linux/mount.h>
@@ -27,7 +28,9 @@
  */
 #define VIRTIO_BALLOON_PAGES_PER_PAGE (unsigned)(PAGE_SIZE >> V= IRTIO_BALLOON_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

 #define VIRTIO_BALLOON_FREE_PAGE_ALLOC_FLAG (__GFP_NORETRY | __GFP_NO= WARN | \
                     = ;                     &nb= sp;  __GFP_NOMEMALLOC)
@@ -112,8 +115,11 @@ struct virtio_balloon {
        /* Memory statistics */
        struct virtio_balloon_stat stats[VIRTIO_BALLOON= _S_NR];

-       /* To register a shrinker to shrink memory upon= memory pressure */
+       /* Shrinker to return free pages - VIRTIO_BALLO= ON_F_FREE_PAGE_HINT */
        struct shrinker shrinker;
+
+       /* OOM notifier to deflate on OOM - VIRTIO_BALL= OON_F_DEFLATE_ON_OOM */
+       struct notifier_block oom_nb;
 };

 static struct virtio_device_id id_table[] =3D {
@@ -786,50 +792,13 @@ static unsigned long shrink_free_pages(struct virtio_= balloon *vb,
        return blocks_freed * VIRTIO_BALLOON_HINT_BLOCK= _PAGES;
 }

-static unsigned long leak_balloon_pages(struct virtio_balloon *vb,
-                    &nbs= p;                     un= signed long pages_to_free)
-{
-       return leak_balloon(vb, pages_to_free * VIRTIO_= BALLOON_PAGES_PER_PAGE) /
-               VIRTIO_BALLOON_PAGE= S_PER_PAGE;
-}
-
-static unsigned long shrink_balloon_pages(struct virtio_balloon *vb,
-                    &nbs= p;                    uns= igned long pages_to_free)
-{
-       unsigned long pages_freed =3D 0;
-
-       /*
-        * One invocation of leak_balloon can deflate a= t most
-        * VIRTIO_BALLOON_ARRAY_PFNS_MAX balloon pages,= so we call it
-        * multiple times to deflate pages till reachin= g pages_to_free.
-        */
-       while (vb->num_pages && pages_freed = < pages_to_free)
-               pages_freed +=3D le= ak_balloon_pages(vb,
-                    &nbs= p;                     &n= bsp;      pages_to_free - pages_freed);
-
-       update_balloon_size(vb);
-
-       return pages_freed;
-}
-
 static unsigned long virtio_balloon_shrinker_scan(struct shrinker *sh= rinker,
                     = ;                     &nb= sp;       struct shrink_control *sc)
 {
-       unsigned long pages_to_free, pages_freed =3D 0;=
        struct virtio_balloon *vb =3D container_of(shri= nker,
                     = ;                   struct vir= tio_balloon, shrinker);

-       pages_to_free =3D sc->nr_to_scan;
-
-       if (virtio_has_feature(vb->vdev, VIRTIO_BALL= OON_F_FREE_PAGE_HINT))
-               pages_freed =3D shr= ink_free_pages(vb, pages_to_free);
-
-       if (pages_freed >=3D pages_to_free)
-               return pages_freed;=
-
-       pages_freed +=3D shrink_balloon_pages(vb, pages= _to_free - pages_freed);
-
-       return pages_freed;
+       return shrink_free_pages(vb, sc->nr_to_scan)= ;
 }

 static unsigned long virtio_balloon_shrinker_count(struct shrinker *s= hrinker,
@@ -837,26 +806,22 @@ static unsigned long virtio_balloon_shrinker_count(st= ruct shrinker *shrinker,
 {
        struct virtio_balloon *vb =3D container_of(shri= nker,
                     = ;                   struct vir= tio_balloon, shrinker);
-       unsigned long count;
-
-       count =3D vb->num_pages / VIRTIO_BALLOON_PAG= ES_PER_PAGE;
-       count +=3D vb->num_free_page_blocks * VIRTIO= _BALLOON_HINT_BLOCK_PAGES;

-       return count;
+       return vb->num_free_page_blocks * VIRTIO_BAL= LOON_HINT_BLOCK_PAGES;
 }

-static void virtio_balloon_unregister_shrinker(struct virtio_balloon *vb)<= br> +static int virtio_balloon_oom_notify(struct notifier_block *nb,
+                    &nbs= p;               unsigned long dummy, vo= id *parm)
 {
-       unregister_shrinker(&vb->shrinker);
-}
+       struct virtio_balloon *vb =3D container_of(nb,<= br> +                    &nbs= p;                     &n= bsp;     struct virtio_balloon, oom_nb);
+       unsigned long *freed =3D parm;

-static int virtio_balloon_register_shrinker(struct virtio_balloon *vb)
-{
-       vb->shrinker.scan_objects =3D virtio_balloon= _shrinker_scan;
-       vb->shrinker.count_objects =3D virtio_balloo= n_shrinker_count;
-       vb->shrinker.seeks =3D DEFAULT_SEEKS;
+       *freed +=3D leak_balloon(vb, VIRTIO_BALLOON_OOM= _NR_PAGES) /
+                 VIRTIO_BALLO= ON_PAGES_PER_PAGE;
+       update_balloon_size(vb);

-       return register_shrinker(&vb->shrinker);=
+       return NOTIFY_OK;
 }

 static int virtballoon_probe(struct virtio_device *vdev)
@@ -933,22 +898,35 @@ static int virtballoon_probe(struct virtio_device *vd= ev)
                     = ;   virtio_cwrite(vb->vdev, struct virtio_balloon_config,
                     = ;                 poison_val, &= poison_val);
                }
-       }
-       /*
-        * We continue to use VIRTIO_BALLOON_F_DEFLATE_= ON_OOM to decide if a
-        * shrinker needs to be registered to relieve m= emory pressure.
-        */
-       if (virtio_has_feature(vb->vdev, VIRTIO_BALL= OON_F_DEFLATE_ON_OOM)) {
-               err =3D virtio_ball= oon_register_shrinker(vb);
+
+               /*
+                * We're allowed to= reuse any free pages, even if they are
+                * still to be proc= essed by the host.
It is important to clarify that pag= es that are on the inflate queue but not ACKed by the host (the queue entry= has not been returned) are _not_ okay to reuse.
If the host is g= oing to do something destructive to the page (like deback it) then that nee= ds to happen before the entry is returned.

While you are correct, this comment is in the = =E2=80=9Efree page hinting=E2=80=9C section/if (not obvious by looking at t= he diff only), so it does not apply to inflate/deflate queues - but only fr= ee pages that are getting hinted. Or am I misreading your suggestion/missin= g something?

Thanks!


=
--Apple-Mail-F9A6D605-018C-4510-9EDF-244179F93224--