Displaying 20 results from an estimated 323 matches for "leak_balloon".
2017 Oct 13
2
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
On Tue, Oct 10, 2017 at 07:47:37PM +0900, Tetsuo Handa wrote:
> In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> serialize against fill_balloon(). But in fill_balloon(),
> alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
> implies __GFP_...
2017 Oct 13
2
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
On Tue, Oct 10, 2017 at 07:47:37PM +0900, Tetsuo Handa wrote:
> In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> serialize against fill_balloon(). But in fill_balloon(),
> alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
> implies __GFP_...
2017 Oct 15
2
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
On Sat, Oct 14, 2017 at 01:41:14AM +0900, Tetsuo Handa wrote:
> Michael S. Tsirkin wrote:
> > On Tue, Oct 10, 2017 at 07:47:37PM +0900, Tetsuo Handa wrote:
> > > In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> > > serialize against fill_balloon(). But in fill_balloon(),
> > > alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> > > called with vb->balloon_lock mutex held. Since GFP_HIGHUSER...
2017 Oct 15
2
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
On Sat, Oct 14, 2017 at 01:41:14AM +0900, Tetsuo Handa wrote:
> Michael S. Tsirkin wrote:
> > On Tue, Oct 10, 2017 at 07:47:37PM +0900, Tetsuo Handa wrote:
> > > In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> > > serialize against fill_balloon(). But in fill_balloon(),
> > > alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> > > called with vb->balloon_lock mutex held. Since GFP_HIGHUSER...
2017 Sep 11
6
mm, virtio: possible OOM lockup at virtballoon_oom_notify()
Hello.
I noticed that virtio_balloon is using register_oom_notifier() and
leak_balloon() from virtballoon_oom_notify() might depend on
__GFP_DIRECT_RECLAIM memory allocation.
In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
serialize against fill_balloon(). But in fill_balloon(),
alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
ca...
2017 Sep 11
6
mm, virtio: possible OOM lockup at virtballoon_oom_notify()
Hello.
I noticed that virtio_balloon is using register_oom_notifier() and
leak_balloon() from virtballoon_oom_notify() might depend on
__GFP_DIRECT_RECLAIM memory allocation.
In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
serialize against fill_balloon(). But in fill_balloon(),
alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
ca...
2017 Oct 12
0
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
On 10/10/2017 06:47 PM, Tetsuo Handa wrote:
> In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> serialize against fill_balloon(). But in fill_balloon(),
> alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
> implies __GFP_...
2017 Oct 12
0
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
On 10/10/2017 06:47 PM, Tetsuo Handa wrote:
> In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> serialize against fill_balloon(). But in fill_balloon(),
> alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
> implies __GFP_...
2017 Oct 22
2
[PATCH v1 1/3] virtio-balloon: replace the coarse-grained balloon_lock
Wei Wang wrote:
> The balloon_lock was used to synchronize the access demand to elements
> of struct virtio_balloon and its queue operations (please see commit
> e22504296d). This prevents the concurrent run of the leak_balloon and
> fill_balloon functions, thereby resulting in a deadlock issue on OOM:
>
> fill_balloon: take balloon_lock and wait for OOM to get some memory;
> oom_notify: release some inflated memory via leak_balloon();
> leak_balloon: wait for balloon_lock to be released by fill_balloon.
&...
2017 Oct 22
2
[PATCH v1 1/3] virtio-balloon: replace the coarse-grained balloon_lock
Wei Wang wrote:
> The balloon_lock was used to synchronize the access demand to elements
> of struct virtio_balloon and its queue operations (please see commit
> e22504296d). This prevents the concurrent run of the leak_balloon and
> fill_balloon functions, thereby resulting in a deadlock issue on OOM:
>
> fill_balloon: take balloon_lock and wait for OOM to get some memory;
> oom_notify: release some inflated memory via leak_balloon();
> leak_balloon: wait for balloon_lock to be released by fill_balloon.
&...
2017 Oct 13
0
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
Michael S. Tsirkin wrote:
> On Tue, Oct 10, 2017 at 07:47:37PM +0900, Tetsuo Handa wrote:
> > In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> > serialize against fill_balloon(). But in fill_balloon(),
> > alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> > called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
>...
2017 Oct 10
0
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
On Tue 10-10-17 19:47:37, Tetsuo Handa wrote:
> In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> serialize against fill_balloon(). But in fill_balloon(),
> alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
> implies __GFP_...
2017 Oct 10
0
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
On Tue 10-10-17 19:47:37, Tetsuo Handa wrote:
> In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> serialize against fill_balloon(). But in fill_balloon(),
> alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
> implies __GFP_...
2017 Oct 13
0
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
...--------
> >From 210dba24134e54cd470e79712c5cb8bb255566c0 Mon Sep 17 00:00:00 2001
> From: Tetsuo Handa <penguin-kernel at I-love.SAKURA.ne.jp>
> Date: Tue, 10 Oct 2017 19:28:20 +0900
> Subject: [PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
>
> In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> serialize against fill_balloon(). But in fill_balloon(),
> alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
> implies __GFP_...
2017 Oct 13
0
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
...--------
> >From 210dba24134e54cd470e79712c5cb8bb255566c0 Mon Sep 17 00:00:00 2001
> From: Tetsuo Handa <penguin-kernel at I-love.SAKURA.ne.jp>
> Date: Tue, 10 Oct 2017 19:28:20 +0900
> Subject: [PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
>
> In leak_balloon(), mutex_lock(&vb->balloon_lock) is called in order to
> serialize against fill_balloon(). But in fill_balloon(),
> alloc_page(GFP_HIGHUSER[_MOVABLE] | __GFP_NOMEMALLOC | __GFP_NORETRY) is
> called with vb->balloon_lock mutex held. Since GFP_HIGHUSER[_MOVABLE]
> implies __GFP_...
2017 Oct 15
0
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
...S. Tsirkin wrote:
> > >
> > > The proper fix isn't that hard - just avoid allocations under lock.
> > >
> > > Patch posted, pls take a look.
> >
> > Your patch allocates pages in order to inflate the balloon, but
> > your patch will allow leak_balloon() to deflate the balloon.
> > How deflating the balloon (i.e. calling leak_balloon()) makes sense
> > when allocating pages for inflating the balloon (i.e. calling
> > fill_balloon()) ?
>
> The idea is that fill_balloon is allocating memory with __GFP_NORETRY
> so it wil...
2017 Oct 22
0
[PATCH v1 1/3] virtio-balloon: replace the coarse-grained balloon_lock
On 10/22/2017 01:20 PM, Tetsuo Handa wrote:
> Wei Wang wrote:
>> The balloon_lock was used to synchronize the access demand to elements
>> of struct virtio_balloon and its queue operations (please see commit
>> e22504296d). This prevents the concurrent run of the leak_balloon and
>> fill_balloon functions, thereby resulting in a deadlock issue on OOM:
>>
>> fill_balloon: take balloon_lock and wait for OOM to get some memory;
>> oom_notify: release some inflated memory via leak_balloon();
>> leak_balloon: wait for balloon_lock to be released...
2014 Oct 15
0
[PATCH 1/2] virtio_balloon: return the amount of freed memory from leak_balloon()
...-)
diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_balloon.c
index f893148..66cac10 100644
--- a/drivers/virtio/virtio_balloon.c
+++ b/drivers/virtio/virtio_balloon.c
@@ -168,8 +168,9 @@ static void release_pages_by_pfn(const u32 pfns[], unsigned int num)
}
}
-static void leak_balloon(struct virtio_balloon *vb, size_t num)
+static unsigned leak_balloon(struct virtio_balloon *vb, size_t num)
{
+ unsigned num_freed_pages;
struct page *page;
struct balloon_dev_info *vb_dev_info = &vb->vb_dev_info;
@@ -186,6 +187,7 @@ static void leak_balloon(struct virtio_balloon *vb,...
2017 Oct 16
2
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
...t;
> > > > The proper fix isn't that hard - just avoid allocations under lock.
> > > >
> > > > Patch posted, pls take a look.
> > >
> > > Your patch allocates pages in order to inflate the balloon, but
> > > your patch will allow leak_balloon() to deflate the balloon.
> > > How deflating the balloon (i.e. calling leak_balloon()) makes sense
> > > when allocating pages for inflating the balloon (i.e. calling
> > > fill_balloon()) ?
> >
> > The idea is that fill_balloon is allocating memory with __G...
2017 Oct 16
2
[PATCH] virtio: avoid possible OOM lockup at virtballoon_oom_notify()
...t;
> > > > The proper fix isn't that hard - just avoid allocations under lock.
> > > >
> > > > Patch posted, pls take a look.
> > >
> > > Your patch allocates pages in order to inflate the balloon, but
> > > your patch will allow leak_balloon() to deflate the balloon.
> > > How deflating the balloon (i.e. calling leak_balloon()) makes sense
> > > when allocating pages for inflating the balloon (i.e. calling
> > > fill_balloon()) ?
> >
> > The idea is that fill_balloon is allocating memory with __G...