* [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects
[not found] <20151124231738.GA15770@nuc-i3427.alporthouse.com>
@ 2015-11-25 18:36 ` Chris Wilson
2015-11-25 18:53 ` Chris Wilson
2015-11-25 19:06 ` Johannes Weiner
0 siblings, 2 replies; 7+ messages in thread
From: Chris Wilson @ 2015-11-25 18:36 UTC (permalink / raw
To: intel-gfx; +Cc: Chris Wilson, linux-mm, Akash Goel, sourab.gupta
If the system has no available swap pages, we cannot make forward
progress in the shrinker by releasing active pages, only by releasing
purgeable pages which are immediately reaped. Take total_swap_pages into
account when counting up available objects to be shrunk and subsequently
shrinking them. By doing so, we avoid unbinding objects that cannot be
shrunk and so wasting CPU cycles flushing those objects from the GPU to
the system and then immediately back again (as they will more than
likely be reused shortly after).
Based on a patch by Akash Goel.
v2: Check for frontswap without physical swap (or dedicated swap space).
If frontswap is available, we may be able to compress the GPU pages
instead of swapping out to disk. In this case, we do want to shrink GPU
objects and so make them available for compressing.
Reported-by: Akash Goel <akash.goel@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: linux-mm@kvack.org
Cc: Akash Goel <akash.goel@intel.com>
Cc: sourab.gupta@intel.com
---
drivers/gpu/drm/i915/i915_gem_shrinker.c | 61 +++++++++++++++++++++++---------
1 file changed, 45 insertions(+), 16 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_gem_shrinker.c b/drivers/gpu/drm/i915/i915_gem_shrinker.c
index f7df54a8ee2b..451a75a056da 100644
--- a/drivers/gpu/drm/i915/i915_gem_shrinker.c
+++ b/drivers/gpu/drm/i915/i915_gem_shrinker.c
@@ -22,6 +22,7 @@
*
*/
+#include <linux/frontswap.h>
#include <linux/oom.h>
#include <linux/shmem_fs.h>
#include <linux/slab.h>
@@ -47,6 +48,46 @@ static bool mutex_is_locked_by(struct mutex *mutex, struct task_struct *task)
#endif
}
+static int num_vma_bound(struct drm_i915_gem_object *obj)
+{
+ struct i915_vma *vma;
+ int count = 0;
+
+ list_for_each_entry(vma, &obj->vma_list, vma_link) {
+ if (drm_mm_node_allocated(&vma->node))
+ count++;
+ if (vma->pin_count)
+ count++;
+ }
+
+ return count;
+}
+
+static bool swap_available(void)
+{
+ return total_swap_pages || frontswap_enabled;
+}
+
+static bool can_release_pages(struct drm_i915_gem_object *obj)
+{
+ /* Only report true if by unbinding the object and putting its pages
+ * we can actually make forward progress towards freeing physical
+ * pages.
+ *
+ * If the pages are pinned for any other reason than being bound
+ * to the GPU, simply unbinding from the GPU is not going to succeed
+ * in release our pin count on the pages themselves.
+ */
+ if (obj->pages_pin_count != num_vma_bound(obj))
+ return false;
+
+ /* We can only return physical pages if we either discard them
+ * (because the user has marked them as being purgeable) or if
+ * we can move their contents out to swap.
+ */
+ return swap_available() || obj->madv == I915_MADV_DONTNEED;
+}
+
/**
* i915_gem_shrink - Shrink buffer object caches
* @dev_priv: i915 device
@@ -129,6 +170,9 @@ i915_gem_shrink(struct drm_i915_private *dev_priv,
if ((flags & I915_SHRINK_ACTIVE) == 0 && obj->active)
continue;
+ if (!can_release_pages(obj))
+ continue;
+
drm_gem_object_reference(&obj->base);
/* For the unbound phase, this should be a no-op! */
@@ -188,21 +232,6 @@ static bool i915_gem_shrinker_lock(struct drm_device *dev, bool *unlock)
return true;
}
-static int num_vma_bound(struct drm_i915_gem_object *obj)
-{
- struct i915_vma *vma;
- int count = 0;
-
- list_for_each_entry(vma, &obj->vma_list, vma_link) {
- if (drm_mm_node_allocated(&vma->node))
- count++;
- if (vma->pin_count)
- count++;
- }
-
- return count;
-}
-
static unsigned long
i915_gem_shrinker_count(struct shrinker *shrinker, struct shrink_control *sc)
{
@@ -222,7 +251,7 @@ i915_gem_shrinker_count(struct shrinker *shrinker, struct shrink_control *sc)
count += obj->base.size >> PAGE_SHIFT;
list_for_each_entry(obj, &dev_priv->mm.bound_list, global_list) {
- if (!obj->active && obj->pages_pin_count == num_vma_bound(obj))
+ if (!obj->active && can_release_pages(obj))
count += obj->base.size >> PAGE_SHIFT;
}
--
2.6.2
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects
2015-11-25 18:36 ` [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects Chris Wilson
@ 2015-11-25 18:53 ` Chris Wilson
2015-11-25 19:06 ` Johannes Weiner
1 sibling, 0 replies; 7+ messages in thread
From: Chris Wilson @ 2015-11-25 18:53 UTC (permalink / raw
To: intel-gfx; +Cc: linux-mm, Akash Goel, sourab.gupta
On Wed, Nov 25, 2015 at 06:36:56PM +0000, Chris Wilson wrote:
> +static bool swap_available(void)
> +{
> + return total_swap_pages || frontswap_enabled;
> +}
Of course these aren't exported symbols, so really this is just RFC
right now.
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects
2015-11-25 18:36 ` [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects Chris Wilson
2015-11-25 18:53 ` Chris Wilson
@ 2015-11-25 19:06 ` Johannes Weiner
2015-11-25 20:31 ` Chris Wilson
1 sibling, 1 reply; 7+ messages in thread
From: Johannes Weiner @ 2015-11-25 19:06 UTC (permalink / raw
To: Chris Wilson; +Cc: intel-gfx, linux-mm, Akash Goel, sourab.gupta
On Wed, Nov 25, 2015 at 06:36:56PM +0000, Chris Wilson wrote:
> If the system has no available swap pages, we cannot make forward
> progress in the shrinker by releasing active pages, only by releasing
> purgeable pages which are immediately reaped. Take total_swap_pages into
> account when counting up available objects to be shrunk and subsequently
> shrinking them. By doing so, we avoid unbinding objects that cannot be
> shrunk and so wasting CPU cycles flushing those objects from the GPU to
> the system and then immediately back again (as they will more than
> likely be reused shortly after).
>
> Based on a patch by Akash Goel.
>
> v2: Check for frontswap without physical swap (or dedicated swap space).
> If frontswap is available, we may be able to compress the GPU pages
> instead of swapping out to disk. In this case, we do want to shrink GPU
> objects and so make them available for compressing.
Frontswap always sits on top of an active swap device. It's enough to
check for available swap space.
> +static bool swap_available(void)
> +{
> + return total_swap_pages || frontswap_enabled;
> +}
If you use get_nr_swap_pages() instead of total_swap_pages, this will
also stop scanning objects once the swap space is full. We do that in
the VM to stop scanning anonymous pages.
On a sidenote, frontswap_enabled is #defined to 1 when the feature is
compiled in, so this would be a no-op on most distro kernels.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects
2015-11-25 19:06 ` Johannes Weiner
@ 2015-11-25 20:31 ` Chris Wilson
2015-11-25 20:46 ` Johannes Weiner
0 siblings, 1 reply; 7+ messages in thread
From: Chris Wilson @ 2015-11-25 20:31 UTC (permalink / raw
To: Johannes Weiner; +Cc: intel-gfx, linux-mm, Akash Goel, sourab.gupta
On Wed, Nov 25, 2015 at 02:06:10PM -0500, Johannes Weiner wrote:
> On Wed, Nov 25, 2015 at 06:36:56PM +0000, Chris Wilson wrote:
> > If the system has no available swap pages, we cannot make forward
> > progress in the shrinker by releasing active pages, only by releasing
> > purgeable pages which are immediately reaped. Take total_swap_pages into
> > account when counting up available objects to be shrunk and subsequently
> > shrinking them. By doing so, we avoid unbinding objects that cannot be
> > shrunk and so wasting CPU cycles flushing those objects from the GPU to
> > the system and then immediately back again (as they will more than
> > likely be reused shortly after).
> >
> > Based on a patch by Akash Goel.
> >
> > v2: Check for frontswap without physical swap (or dedicated swap space).
> > If frontswap is available, we may be able to compress the GPU pages
> > instead of swapping out to disk. In this case, we do want to shrink GPU
> > objects and so make them available for compressing.
>
> Frontswap always sits on top of an active swap device. It's enough to
> check for available swap space.
>
> > +static bool swap_available(void)
> > +{
> > + return total_swap_pages || frontswap_enabled;
> > +}
>
> If you use get_nr_swap_pages() instead of total_swap_pages, this will
> also stop scanning objects once the swap space is full. We do that in
> the VM to stop scanning anonymous pages.
Thanks. Would EXPORT_SYMBOL_GPL(nr_swap_pages) (or equivalent) be
acceptable?
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects
2015-11-25 20:31 ` Chris Wilson
@ 2015-11-25 20:46 ` Johannes Weiner
2015-11-26 11:25 ` Chris Wilson
0 siblings, 1 reply; 7+ messages in thread
From: Johannes Weiner @ 2015-11-25 20:46 UTC (permalink / raw
To: Chris Wilson; +Cc: intel-gfx, linux-mm, Akash Goel, sourab.gupta
On Wed, Nov 25, 2015 at 08:31:02PM +0000, Chris Wilson wrote:
> On Wed, Nov 25, 2015 at 02:06:10PM -0500, Johannes Weiner wrote:
> > On Wed, Nov 25, 2015 at 06:36:56PM +0000, Chris Wilson wrote:
> > > +static bool swap_available(void)
> > > +{
> > > + return total_swap_pages || frontswap_enabled;
> > > +}
> >
> > If you use get_nr_swap_pages() instead of total_swap_pages, this will
> > also stop scanning objects once the swap space is full. We do that in
> > the VM to stop scanning anonymous pages.
>
> Thanks. Would EXPORT_SYMBOL_GPL(nr_swap_pages) (or equivalent) be
> acceptable?
No opposition from me. Just please add a small comment that this is
for shrinkers with swappable objects.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects
2015-11-25 20:46 ` Johannes Weiner
@ 2015-11-26 11:25 ` Chris Wilson
2015-11-26 15:40 ` Johannes Weiner
0 siblings, 1 reply; 7+ messages in thread
From: Chris Wilson @ 2015-11-26 11:25 UTC (permalink / raw
To: Johannes Weiner; +Cc: intel-gfx, linux-mm, Akash Goel, sourab.gupta
On Wed, Nov 25, 2015 at 03:46:35PM -0500, Johannes Weiner wrote:
> On Wed, Nov 25, 2015 at 08:31:02PM +0000, Chris Wilson wrote:
> > On Wed, Nov 25, 2015 at 02:06:10PM -0500, Johannes Weiner wrote:
> > > On Wed, Nov 25, 2015 at 06:36:56PM +0000, Chris Wilson wrote:
> > > > +static bool swap_available(void)
> > > > +{
> > > > + return total_swap_pages || frontswap_enabled;
> > > > +}
> > >
> > > If you use get_nr_swap_pages() instead of total_swap_pages, this will
> > > also stop scanning objects once the swap space is full. We do that in
> > > the VM to stop scanning anonymous pages.
> >
> > Thanks. Would EXPORT_SYMBOL_GPL(nr_swap_pages) (or equivalent) be
> > acceptable?
>
> No opposition from me. Just please add a small comment that this is
> for shrinkers with swappable objects.
diff --git a/mm/swapfile.c b/mm/swapfile.c
index 58877312cf6b..1c7861f4c43c 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -48,6 +48,14 @@ static sector_t map_swap_entry(swp_entry_t, struct block_device**);
DEFINE_SPINLOCK(swap_lock);
static unsigned int nr_swapfiles;
atomic_long_t nr_swap_pages;
+/*
+ * Some modules use swappable objects and may try to swap them out under
+ * memory pressure (via the shrinker). Before doing so, they may wish to
+ * check to see if any swap space is available. The shrinker also directly
+ * uses the available swap space to determine whether it can swapout
+ * anon pages in the same manner.
+ */
+EXPORT_SYMBOL_GPL(nr_swap_pages);
Something like that, after a couple more edits?
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects
2015-11-26 11:25 ` Chris Wilson
@ 2015-11-26 15:40 ` Johannes Weiner
0 siblings, 0 replies; 7+ messages in thread
From: Johannes Weiner @ 2015-11-26 15:40 UTC (permalink / raw
To: Chris Wilson; +Cc: intel-gfx, linux-mm, Akash Goel, sourab.gupta
On Thu, Nov 26, 2015 at 11:25:14AM +0000, Chris Wilson wrote:
> On Wed, Nov 25, 2015 at 03:46:35PM -0500, Johannes Weiner wrote:
> > On Wed, Nov 25, 2015 at 08:31:02PM +0000, Chris Wilson wrote:
> > > On Wed, Nov 25, 2015 at 02:06:10PM -0500, Johannes Weiner wrote:
> > > > On Wed, Nov 25, 2015 at 06:36:56PM +0000, Chris Wilson wrote:
> > > > > +static bool swap_available(void)
> > > > > +{
> > > > > + return total_swap_pages || frontswap_enabled;
> > > > > +}
> > > >
> > > > If you use get_nr_swap_pages() instead of total_swap_pages, this will
> > > > also stop scanning objects once the swap space is full. We do that in
> > > > the VM to stop scanning anonymous pages.
> > >
> > > Thanks. Would EXPORT_SYMBOL_GPL(nr_swap_pages) (or equivalent) be
> > > acceptable?
> >
> > No opposition from me. Just please add a small comment that this is
> > for shrinkers with swappable objects.
>
> diff --git a/mm/swapfile.c b/mm/swapfile.c
> index 58877312cf6b..1c7861f4c43c 100644
> --- a/mm/swapfile.c
> +++ b/mm/swapfile.c
> @@ -48,6 +48,14 @@ static sector_t map_swap_entry(swp_entry_t, struct block_device**);
> DEFINE_SPINLOCK(swap_lock);
> static unsigned int nr_swapfiles;
> atomic_long_t nr_swap_pages;
> +/*
> + * Some modules use swappable objects and may try to swap them out under
> + * memory pressure (via the shrinker). Before doing so, they may wish to
> + * check to see if any swap space is available. The shrinker also directly
> + * uses the available swap space to determine whether it can swapout
> + * anon pages in the same manner.
> + */
> +EXPORT_SYMBOL_GPL(nr_swap_pages);
>
> Something like that, after a couple more edits?
The last sentence isn't necessary IMO, but other than that it looks
good to me.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2015-11-26 15:40 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <20151124231738.GA15770@nuc-i3427.alporthouse.com>
2015-11-25 18:36 ` [PATCH v2] drm/i915: Disable shrinker for non-swapped backed objects Chris Wilson
2015-11-25 18:53 ` Chris Wilson
2015-11-25 19:06 ` Johannes Weiner
2015-11-25 20:31 ` Chris Wilson
2015-11-25 20:46 ` Johannes Weiner
2015-11-26 11:25 ` Chris Wilson
2015-11-26 15:40 ` Johannes Weiner
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).