[rh7,v2,3/4] slub: Make cpu partial slab support configurable

Submitted by Andrey Ryabinin on Dec. 29, 2018, 4:22 p.m.

Details

Message ID 20181229162201.4410-3-aryabinin@virtuozzo.com
State New
Series "Series without cover letter"
Headers show

Commit Message

Andrey Ryabinin Dec. 29, 2018, 4:22 p.m.
From: Joonsoo Kim <iamjoonsoo.kim@lge.com>

CPU partial support can introduce level of indeterminism that is not
wanted in certain context (like a realtime kernel). Make it
configurable.

This patch is based on Christoph Lameter's "slub: Make cpu partial slab
support configurable V2".

Acked-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>

https://jira.sw.ru/browse/PSBM-83199
[ Setting SLUB_CPU_PARTIAL=n save us about a 1G in dvd-store test ]
(cherry picked from commit 345c905d13a4ec9f774b6b4bc038fe4aef26cced)
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
---
 init/Kconfig | 11 +++++++++++
 mm/slub.c    | 28 +++++++++++++++++++++-------
 2 files changed, 32 insertions(+), 7 deletions(-)

Patch hide | download patch | download mbox

diff --git a/init/Kconfig b/init/Kconfig
index 0cbeec104e9a..f48c74bcedac 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1597,6 +1597,17 @@  config SLOB
 
 endchoice
 
+config SLUB_CPU_PARTIAL
+	default y
+	depends on SLUB
+	bool "SLUB per cpu partial cache"
+	help
+	  Per cpu partial caches accellerate objects allocation and freeing
+	  that is local to a processor at the price of more indeterminism
+	  in the latency of the free. On overflow these caches will be cleared
+	  which requires the taking of locks that may cause latency spikes.
+	  Typically one would choose no for a realtime system.
+
 config MMAP_ALLOW_UNINITIALIZED
 	bool "Allow mmapped anonymous memory to be uninitialized"
 	depends on EXPERT && !MMU
diff --git a/mm/slub.c b/mm/slub.c
index d5b0bc1fcd56..de39b176c995 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -124,6 +124,15 @@  static inline int kmem_cache_debug(struct kmem_cache *s)
 #endif
 }
 
+static inline bool kmem_cache_has_cpu_partial(struct kmem_cache *s)
+{
+#ifdef CONFIG_SLUB_CPU_PARTIAL
+	return !kmem_cache_debug(s);
+#else
+	return false;
+#endif
+}
+
 /*
  * Issues still to be resolved:
  *
@@ -1685,7 +1694,8 @@  static void *get_partial_node(struct kmem_cache *s, struct kmem_cache_node *n,
 			put_cpu_partial(s, page, 0);
 			stat(s, CPU_PARTIAL_NODE);
 		}
-		if (kmem_cache_debug(s) || available > s->cpu_partial / 2)
+		if (!kmem_cache_has_cpu_partial(s)
+			|| available > s->cpu_partial / 2)
 			break;
 
 	}
@@ -1999,6 +2009,7 @@  static void deactivate_slab(struct kmem_cache *s, struct page *page, void *freel
 static void unfreeze_partials(struct kmem_cache *s,
 		struct kmem_cache_cpu *c)
 {
+#ifdef CONFIG_SLUB_CPU_PARTIAL
 	struct kmem_cache_node *n = NULL, *n2 = NULL;
 	struct page *page, *discard_page = NULL;
 
@@ -2053,6 +2064,7 @@  static void unfreeze_partials(struct kmem_cache *s,
 		discard_slab(s, page);
 		stat(s, FREE_SLAB);
 	}
+#endif
 }
 
 /*
@@ -2066,6 +2078,7 @@  static void unfreeze_partials(struct kmem_cache *s,
  */
 static void put_cpu_partial(struct kmem_cache *s, struct page *page, int drain)
 {
+#ifdef CONFIG_SLUB_CPU_PARTIAL
 	struct page *oldpage;
 	int pages;
 	int pobjects;
@@ -2112,6 +2125,7 @@  static void put_cpu_partial(struct kmem_cache *s, struct page *page, int drain)
 		local_irq_restore(flags);
 	}
 	preempt_enable();
+#endif
 }
 
 static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c)
@@ -2638,8 +2652,7 @@  static void __slab_free(struct kmem_cache *s, struct page *page,
 		new.inuse -= cnt;
 		if ((!new.inuse || !prior) && !was_frozen) {
 
-			if (!kmem_cache_debug(s) && !prior) {
-
+			if (kmem_cache_has_cpu_partial(s) && !prior) {
 				/*
 				 * Slab was on no list before and will be partially empty
 				 * We can defer the list move and instead freeze it.
@@ -2693,8 +2706,9 @@  static void __slab_free(struct kmem_cache *s, struct page *page,
 	 * Objects left in the slab. If it was not on the partial list before
 	 * then add it.
 	 */
-	if (kmem_cache_debug(s) && unlikely(!prior)) {
-		remove_full(s, n, page);
+	if (!kmem_cache_has_cpu_partial(s) && unlikely(!prior)) {
+		if (kmem_cache_debug(s))
+			remove_full(s, n, page);
 		add_partial(n, page, DEACTIVATE_TO_TAIL);
 		stat(s, FREE_ADD_PARTIAL);
 	}
@@ -3396,7 +3410,7 @@  static int kmem_cache_open(struct kmem_cache *s, unsigned long flags)
 	 *    per node list when we run out of per cpu objects. We only fetch 50%
 	 *    to keep some capacity around for frees.
 	 */
-	if (kmem_cache_debug(s))
+	if (!kmem_cache_has_cpu_partial(s))
 		s->cpu_partial = 0;
 	else if (s->size >= PAGE_SIZE)
 		s->cpu_partial = 2;
@@ -4836,7 +4850,7 @@  static ssize_t cpu_partial_store(struct kmem_cache *s, const char *buf,
 	err = strict_strtoul(buf, 10, &objects);
 	if (err)
 		return err;
-	if (objects && kmem_cache_debug(s))
+	if (objects && !kmem_cache_has_cpu_partial(s))
 		return -EINVAL;
 
 	s->cpu_partial = objects;