The defrag_ratio is used to set the threshold at which defragmentation
should be run on a slabcache.
The allocation ratio is measured in a percentage of the available slots.
The percentage will be lower for slabs that are more fragmented.
Add a defrag ratio field and set it to 30% by default. A limit of 30% specified
that less than 3 out of 10 available slots for objects are in use before
reclaim occurs.
Reviewed-by: Rik van Riel <[email protected]>
Signed-off-by: Christoph Lameter <[email protected]>
---
include/linux/slub_def.h | 7 +++++++
mm/slub.c | 18 ++++++++++++++++++
2 files changed, 25 insertions(+)
Index: linux-2.6/include/linux/slub_def.h
===================================================================
--- linux-2.6.orig/include/linux/slub_def.h 2007-11-06 12:36:28.000000000 -0800
+++ linux-2.6/include/linux/slub_def.h 2007-11-06 12:37:44.000000000 -0800
@@ -53,6 +53,13 @@ struct kmem_cache {
void (*ctor)(struct kmem_cache *, void *);
int inuse; /* Offset to metadata */
int align; /* Alignment */
+ int defrag_ratio; /*
+ * objects/possible-objects limit. If we have
+ * less that the specified percentage of
+ * objects allocated then defrag passes
+ * will start to occur during reclaim.
+ */
+
const char *name; /* Name (only for display!) */
struct list_head list; /* List of slab caches */
#ifdef CONFIG_SLUB_DEBUG
Index: linux-2.6/mm/slub.c
===================================================================
--- linux-2.6.orig/mm/slub.c 2007-11-06 12:37:25.000000000 -0800
+++ linux-2.6/mm/slub.c 2007-11-06 12:37:44.000000000 -0800
@@ -2363,6 +2363,7 @@ static int kmem_cache_open(struct kmem_c
goto error;
s->refcount = 1;
+ s->defrag_ratio = 30;
#ifdef CONFIG_NUMA
s->remote_node_defrag_ratio = 100;
#endif
@@ -4005,6 +4006,22 @@ static ssize_t free_calls_show(struct km
}
SLAB_ATTR_RO(free_calls);
+static ssize_t defrag_ratio_show(struct kmem_cache *s, char *buf)
+{
+ return sprintf(buf, "%d\n", s->defrag_ratio);
+}
+
+static ssize_t defrag_ratio_store(struct kmem_cache *s,
+ const char *buf, size_t length)
+{
+ int n = simple_strtoul(buf, NULL, 10);
+
+ if (n < 100)
+ s->defrag_ratio = n;
+ return length;
+}
+SLAB_ATTR(defrag_ratio);
+
#ifdef CONFIG_NUMA
static ssize_t remote_node_defrag_ratio_show(struct kmem_cache *s, char *buf)
{
@@ -4047,6 +4064,7 @@ static struct attribute * slab_attrs[] =
&shrink_attr.attr,
&alloc_calls_attr.attr,
&free_calls_attr.attr,
+ &defrag_ratio_attr.attr,
#ifdef CONFIG_ZONE_DMA
&cache_dma_attr.attr,
#endif
--
Hi Christoph,
On Tue, Nov 06, 2007 at 05:11:37PM -0800, Christoph Lameter wrote:
> --- linux-2.6.orig/include/linux/slub_def.h 2007-11-06 12:36:28.000000000 -0800
> +++ linux-2.6/include/linux/slub_def.h 2007-11-06 12:37:44.000000000 -0800
> @@ -53,6 +53,13 @@ struct kmem_cache {
> void (*ctor)(struct kmem_cache *, void *);
> int inuse; /* Offset to metadata */
> int align; /* Alignment */
> + int defrag_ratio; /*
> + * objects/possible-objects limit. If we have
> + * less that the specified percentage of
That should be `less than', I guess.
Hannes
On Wed, 7 Nov 2007, Johannes Weiner wrote:
> Hi Christoph,
>
> On Tue, Nov 06, 2007 at 05:11:37PM -0800, Christoph Lameter wrote:
> > --- linux-2.6.orig/include/linux/slub_def.h 2007-11-06 12:36:28.000000000 -0800
> > +++ linux-2.6/include/linux/slub_def.h 2007-11-06 12:37:44.000000000 -0800
> > @@ -53,6 +53,13 @@ struct kmem_cache {
> > void (*ctor)(struct kmem_cache *, void *);
> > int inuse; /* Offset to metadata */
> > int align; /* Alignment */
> > + int defrag_ratio; /*
> > + * objects/possible-objects limit. If we have
> > + * less that the specified percentage of
>
> That should be `less than', I guess.
Correct.
On (06/11/07 17:11), Christoph Lameter didst pronounce:
> The defrag_ratio is used to set the threshold at which defragmentation
> should be run on a slabcache.
>
I'm thick, I would like to see a quick note here on what defragmentation
means. Also, this defrag_ratio seems to have a significantly different
meaning to the other defrag_ratio which isn't helping my poor head at
all.
"The defrag_ratio sets a threshold at which a slab will be vacated of all
it's objects and the pages freed during memory reclaim."
?
> The allocation ratio is measured in a percentage of the available slots.
> The percentage will be lower for slabs that are more fragmented.
>
> Add a defrag ratio field and set it to 30% by default. A limit of 30% specified
> that less than 3 out of 10 available slots for objects are in use before
> reclaim occurs.
>
> Reviewed-by: Rik van Riel <[email protected]>
> Signed-off-by: Christoph Lameter <[email protected]>
> ---
> include/linux/slub_def.h | 7 +++++++
> mm/slub.c | 18 ++++++++++++++++++
> 2 files changed, 25 insertions(+)
>
> Index: linux-2.6/include/linux/slub_def.h
> ===================================================================
> --- linux-2.6.orig/include/linux/slub_def.h 2007-11-06 12:36:28.000000000 -0800
> +++ linux-2.6/include/linux/slub_def.h 2007-11-06 12:37:44.000000000 -0800
> @@ -53,6 +53,13 @@ struct kmem_cache {
> void (*ctor)(struct kmem_cache *, void *);
> int inuse; /* Offset to metadata */
> int align; /* Alignment */
> + int defrag_ratio; /*
> + * objects/possible-objects limit. If we have
> + * less that the specified percentage of
> + * objects allocated then defrag passes
> + * will start to occur during reclaim.
> + */
> +
> const char *name; /* Name (only for display!) */
> struct list_head list; /* List of slab caches */
> #ifdef CONFIG_SLUB_DEBUG
> Index: linux-2.6/mm/slub.c
> ===================================================================
> --- linux-2.6.orig/mm/slub.c 2007-11-06 12:37:25.000000000 -0800
> +++ linux-2.6/mm/slub.c 2007-11-06 12:37:44.000000000 -0800
> @@ -2363,6 +2363,7 @@ static int kmem_cache_open(struct kmem_c
> goto error;
>
> s->refcount = 1;
> + s->defrag_ratio = 30;
> #ifdef CONFIG_NUMA
> s->remote_node_defrag_ratio = 100;
> #endif
> @@ -4005,6 +4006,22 @@ static ssize_t free_calls_show(struct km
> }
> SLAB_ATTR_RO(free_calls);
>
> +static ssize_t defrag_ratio_show(struct kmem_cache *s, char *buf)
> +{
> + return sprintf(buf, "%d\n", s->defrag_ratio);
> +}
> +
> +static ssize_t defrag_ratio_store(struct kmem_cache *s,
> + const char *buf, size_t length)
> +{
> + int n = simple_strtoul(buf, NULL, 10);
> +
> + if (n < 100)
> + s->defrag_ratio = n;
> + return length;
> +}
> +SLAB_ATTR(defrag_ratio);
> +
> #ifdef CONFIG_NUMA
> static ssize_t remote_node_defrag_ratio_show(struct kmem_cache *s, char *buf)
> {
> @@ -4047,6 +4064,7 @@ static struct attribute * slab_attrs[] =
> &shrink_attr.attr,
> &alloc_calls_attr.attr,
> &free_calls_attr.attr,
> + &defrag_ratio_attr.attr,
> #ifdef CONFIG_ZONE_DMA
> &cache_dma_attr.attr,
> #endif
>
> --
>
--
--
Mel Gorman
Part-time Phd Student Linux Technology Center
University of Limerick IBM Dublin Software Lab
On Thu, 8 Nov 2007, Mel Gorman wrote:
> On (06/11/07 17:11), Christoph Lameter didst pronounce:
> > The defrag_ratio is used to set the threshold at which defragmentation
> > should be run on a slabcache.
> >
>
> I'm thick, I would like to see a quick note here on what defragmentation
> means. Also, this defrag_ratio seems to have a significantly different
> meaning to the other defrag_ratio which isn't helping my poor head at
> all.
Yes that is why they have different names. The remote_node_defrag ratio
controls the amount of remote allocs we do to reduce fragmentation.
> "The defrag_ratio sets a threshold at which a slab will be vacated of all
> it's objects and the pages freed during memory reclaim."
Sortof. If a slab is beyond the threshold during reclaim then reclaim will
attempt to free the remaining objects in the slab to reclaim the whole
slab.