SLUB: dynamic per-cache MIN_PARTIAL

This patch changes the static MIN_PARTIAL to a dynamic per-cache ->min_partial
value that is calculated from object size. The bigger the object size, the more
pages we keep on the partial list.

I tested SLAB, SLUB, and SLUB with this patch on Jens Axboe's 'netio' example
script of the fio benchmarking tool. The script stresses the networking
subsystem which should also give a fairly good beating of kmalloc() et al.

To run the test yourself, first clone the fio repository:

  git clone git://git.kernel.dk/fio.git

and then run the following command n times on your machine:

  time ./fio examples/netio

The results on my 2-way 64-bit x86 machine are as follows:

  [ the minimum, maximum, and average are captured from 50 individual runs ]

                 real time (seconds)
                 min      max      avg      sd
  SLAB           22.76    23.38    22.98    0.17
  SLUB           22.80    25.78    23.46    0.72
  SLUB (dynamic) 22.74    23.54    23.00    0.20

                 sys time (seconds)
                 min      max      avg      sd
  SLAB           6.90     8.28     7.70     0.28
  SLUB           7.42     16.95    8.89     2.28
  SLUB (dynamic) 7.17     8.64     7.73     0.29

                 user time (seconds)
                 min      max      avg      sd
  SLAB           36.89    38.11    37.50    0.29
  SLUB           30.85    37.99    37.06    1.67
  SLUB (dynamic) 36.75    38.07    37.59    0.32

As you can see from the above numbers, this patch brings SLUB to the same level
as SLAB for this particular workload fixing a ~2% regression. I'd expect this
change to help similar workloads that allocate a lot of objects that are close
to the size of a page.

Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>

diff --git a/include/linux/slub_def.h b/include/linux/slub_def.h
index 5bad61a93f65566cc59716ebe4cae2e0d5f88f14..2f5c16b1aacd3d7bd83a50d90fcba527017a0373 100644 (file)
--- a/include/linux/slub_def.h
+++ b/include/linux/slub_def.h
@@ -46,6 +46,7 @@ struct kmem_cache_cpu {
 struct kmem_cache_node {
        spinlock_t list_lock;   /* Protect partial list and nr_partial */
        unsigned long nr_partial;
+       unsigned long min_partial;
        struct list_head partial;
 #ifdef CONFIG_SLUB_DEBUG
        atomic_long_t nr_slabs;

diff --git a/mm/slub.c b/mm/slub.c
index c26d4c36fba9bbafde6d258aa84d3e3f284dc33a..4f5b96149458a6712d1592fd64bdf4a4b876f38a 100644 (file)
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -1329,7 +1329,7 @@ static struct page *get_any_partial(struct kmem_cache *s, gfp_t flags)
                n = get_node(s, zone_to_nid(zone));
 
                if (n && cpuset_zone_allowed_hardwall(zone, flags) &&
-                               n->nr_partial > MIN_PARTIAL) {
+                               n->nr_partial > n->min_partial) {
                        page = get_partial_node(n);
                        if (page)
                                return page;
@@ -1381,7 +1381,7 @@ static void unfreeze_slab(struct kmem_cache *s, struct page *page, int tail)
                slab_unlock(page);
        } else {
                stat(c, DEACTIVATE_EMPTY);
-               if (n->nr_partial < MIN_PARTIAL) {
+               if (n->nr_partial < n->min_partial) {
                        /*
                         * Adding an empty slab to the partial slabs in order
                         * to avoid page allocator overhead. This slab needs
@@ -1913,9 +1913,21 @@ static void init_kmem_cache_cpu(struct kmem_cache *s,
 #endif
 }
 
-static void init_kmem_cache_node(struct kmem_cache_node *n)
+static void
+init_kmem_cache_node(struct kmem_cache_node *n, struct kmem_cache *s)
 {
        n->nr_partial = 0;
+
+       /*
+        * The larger the object size is, the more pages we want on the partial
+        * list to avoid pounding the page allocator excessively.
+        */
+       n->min_partial = ilog2(s->size);
+       if (n->min_partial < MIN_PARTIAL)
+               n->min_partial = MIN_PARTIAL;
+       else if (n->min_partial > MAX_PARTIAL)
+               n->min_partial = MAX_PARTIAL;
+
        spin_lock_init(&n->list_lock);
        INIT_LIST_HEAD(&n->partial);
 #ifdef CONFIG_SLUB_DEBUG
@@ -2087,7 +2099,7 @@ static struct kmem_cache_node *early_kmem_cache_node_alloc(gfp_t gfpflags,
        init_object(kmalloc_caches, n, 1);
        init_tracking(kmalloc_caches, n);
 #endif
-       init_kmem_cache_node(n);
+       init_kmem_cache_node(n, kmalloc_caches);
        inc_slabs_node(kmalloc_caches, node, page->objects);
 
        /*
@@ -2144,7 +2156,7 @@ static int init_kmem_cache_nodes(struct kmem_cache *s, gfp_t gfpflags)
 
                }
                s->node[node] = n;
-               init_kmem_cache_node(n);
+               init_kmem_cache_node(n, s);
        }
        return 1;
 }
@@ -2155,7 +2167,7 @@ static void free_kmem_cache_nodes(struct kmem_cache *s)
 
 static int init_kmem_cache_nodes(struct kmem_cache *s, gfp_t gfpflags)
 {
-       init_kmem_cache_node(&s->local_node);
+       init_kmem_cache_node(&s->local_node, s);
        return 1;
 }
 #endif
@@ -2889,7 +2901,7 @@ static int slab_mem_going_online_callback(void *arg)
                        ret = -ENOMEM;
                        goto out;
                }
-               init_kmem_cache_node(n);
+               init_kmem_cache_node(n, s);
                s->node[nid] = n;
        }
 out: