2 files changed, 207 insertions, 3 deletions

diff --git a/include/linux/slub_def.h b/include/linux/slub_def.h
index 875df55ab36d..009b0020079d 100644
--- a/include/linux/slub_def.h
+++ b/include/linux/slub_def.h
@@ -35,7 +35,10 @@ enum stat_item {
         NR_SLUB_STAT_ITEMS };
 
 struct kmem_cache_cpu {
-        void **freelist;        /* Pointer to first free per cpu object */
+        void **freelist;        /* Pointer to next available object */
+#ifdef CONFIG_CMPXCHG_LOCAL
+        unsigned long tid;      /* Globally unique transaction id */
+#endif
         struct page *page;      /* The slab from which we are allocating */
         int node;               /* The node of the page (or -1 for debug) */
 #ifdef CONFIG_SLUB_STATS
diff --git a/mm/slub.c b/mm/slub.c
index bae7a5c636f4..65030c7fd7e2 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -1494,6 +1494,77 @@ static void unfreeze_slab(struct kmem_cache *s, struct page *page, int tail)
         }
 }
 
+#ifdef CONFIG_CMPXCHG_LOCAL
+#ifdef CONFIG_PREEMPT
+/*
+ * Calculate the next globally unique transaction for disambiguiation
+ * during cmpxchg. The transactions start with the cpu number and are then
+ * incremented by CONFIG_NR_CPUS.
+ */
+#define TID_STEP  roundup_pow_of_two(CONFIG_NR_CPUS)
+#else
+/*
+ * No preemption supported therefore also no need to check for
+ * different cpus.
+ */
+#define TID_STEP 1
+#endif
+
+static inline unsigned long next_tid(unsigned long tid)
+{
+        return tid + TID_STEP;
+}
+
+static inline unsigned int tid_to_cpu(unsigned long tid)
+{
+        return tid % TID_STEP;
+}
+
+static inline unsigned long tid_to_event(unsigned long tid)
+{
+        return tid / TID_STEP;
+}
+
+static inline unsigned int init_tid(int cpu)
+{
+        return cpu;
+}
+
+static inline void note_cmpxchg_failure(const char *n,
+                const struct kmem_cache *s, unsigned long tid)
+{
+#ifdef SLUB_DEBUG_CMPXCHG
+        unsigned long actual_tid = __this_cpu_read(s->cpu_slab->tid);
+
+        printk(KERN_INFO "%s %s: cmpxchg redo ", n, s->name);
+
+#ifdef CONFIG_PREEMPT
+        if (tid_to_cpu(tid) != tid_to_cpu(actual_tid))
+                printk("due to cpu change %d -> %d\n",
+                        tid_to_cpu(tid), tid_to_cpu(actual_tid));
+        else
+#endif
+        if (tid_to_event(tid) != tid_to_event(actual_tid))
+                printk("due to cpu running other code. Event %ld->%ld\n",
+                        tid_to_event(tid), tid_to_event(actual_tid));
+        else
+                printk("for unknown reason: actual=%lx was=%lx target=%lx\n",
+                        actual_tid, tid, next_tid(tid));
+#endif
+}
+
+#endif
+
+void init_kmem_cache_cpus(struct kmem_cache *s)
+{
+#if defined(CONFIG_CMPXCHG_LOCAL) && defined(CONFIG_PREEMPT)
+        int cpu;
+
+        for_each_possible_cpu(cpu)
+                per_cpu_ptr(s->cpu_slab, cpu)->tid = init_tid(cpu);
+#endif
+
+}
 /*
  * Remove the cpu slab
  */
@@ -1525,6 +1596,9 @@ static void deactivate_slab(struct kmem_cache *s, struct kmem_cache_cpu *c)
                 page->inuse--;
         }
         c->page = NULL;
+#ifdef CONFIG_CMPXCHG_LOCAL
+        c->tid = next_tid(c->tid);
+#endif
         unfreeze_slab(s, page, tail);
 }
 
@@ -1659,6 +1733,19 @@ static void *__slab_alloc(struct kmem_cache *s, gfp_t gfpflags, int node,
 {
         void **object;
         struct page *new;
+#ifdef CONFIG_CMPXCHG_LOCAL
+        unsigned long flags;
+
+        local_irq_save(flags);
+#ifdef CONFIG_PREEMPT
+        /*
+         * We may have been preempted and rescheduled on a different
+         * cpu before disabling interrupts. Need to reload cpu area
+         * pointer.
+         */
+        c = this_cpu_ptr(s->cpu_slab);
+#endif
+#endif
 
         /* We handle __GFP_ZERO in the caller */
         gfpflags &= ~__GFP_ZERO;
@@ -1685,6 +1772,10 @@ load_freelist:
         c->node = page_to_nid(c->page);
 unlock_out:
         slab_unlock(c->page);
+#ifdef CONFIG_CMPXCHG_LOCAL
+        c->tid = next_tid(c->tid);
+        local_irq_restore(flags);
+#endif
         stat(s, ALLOC_SLOWPATH);
         return object;
 
@@ -1746,23 +1837,76 @@ static __always_inline void *slab_alloc(struct kmem_cache *s,
 {
         void **object;
         struct kmem_cache_cpu *c;
+#ifdef CONFIG_CMPXCHG_LOCAL
+        unsigned long tid;
+#else
         unsigned long flags;
+#endif
 
         if (slab_pre_alloc_hook(s, gfpflags))
                 return NULL;
 
+#ifndef CONFIG_CMPXCHG_LOCAL
         local_irq_save(flags);
+#else
+redo:
+#endif
+
+        /*
+         * Must read kmem_cache cpu data via this cpu ptr. Preemption is
+         * enabled. We may switch back and forth between cpus while
+         * reading from one cpu area. That does not matter as long
+         * as we end up on the original cpu again when doing the cmpxchg.
+         */
         c = __this_cpu_ptr(s->cpu_slab);
+
+#ifdef CONFIG_CMPXCHG_LOCAL
+        /*
+         * The transaction ids are globally unique per cpu and per operation on
+         * a per cpu queue. Thus they can be guarantee that the cmpxchg_double
+         * occurs on the right processor and that there was no operation on the
+         * linked list in between.
+         */
+        tid = c->tid;
+        barrier();
+#endif
+
         object = c->freelist;
         if (unlikely(!object || !node_match(c, node)))
 
                 object = __slab_alloc(s, gfpflags, node, addr, c);
 
         else {
+#ifdef CONFIG_CMPXCHG_LOCAL
+                /*
+                 * The cmpxchg will only match if there was no additonal
+                 * operation and if we are on the right processor.
+                 *
+                 * The cmpxchg does the following atomically (without lock semantics!)
+                 * 1. Relocate first pointer to the current per cpu area.
+                 * 2. Verify that tid and freelist have not been changed
+                 * 3. If they were not changed replace tid and freelist
+                 *
+                 * Since this is without lock semantics the protection is only against
+                 * code executing on this cpu *not* from access by other cpus.
+                 */
+                if (unlikely(!this_cpu_cmpxchg_double(
+                                s->cpu_slab->freelist, s->cpu_slab->tid,
+                                object, tid,
+                                get_freepointer(s, object), next_tid(tid)))) {
+
+                        note_cmpxchg_failure("slab_alloc", s, tid);
+                        goto redo;
+                }
+#else
                 c->freelist = get_freepointer(s, object);
+#endif
                 stat(s, ALLOC_FASTPATH);
         }
+
+#ifndef CONFIG_CMPXCHG_LOCAL
         local_irq_restore(flags);
+#endif
 
         if (unlikely(gfpflags & __GFP_ZERO) && object)
                 memset(object, 0, s->objsize);
@@ -1840,9 +1984,13 @@ static void __slab_free(struct kmem_cache *s, struct page *page,
 {
         void *prior;
         void **object = (void *)x;
+#ifdef CONFIG_CMPXCHG_LOCAL
+        unsigned long flags;
 
-        stat(s, FREE_SLOWPATH);
+        local_irq_save(flags);
+#endif
         slab_lock(page);
+        stat(s, FREE_SLOWPATH);
 
         if (kmem_cache_debug(s))
                 goto debug;
@@ -1872,6 +2020,9 @@ checks_ok:
 
 out_unlock:
         slab_unlock(page);
+#ifdef CONFIG_CMPXCHG_LOCAL
+        local_irq_restore(flags);
+#endif
         return;
 
 slab_empty:
@@ -1883,6 +2034,9 @@ slab_empty:
                 stat(s, FREE_REMOVE_PARTIAL);
         }
         slab_unlock(page);
+#ifdef CONFIG_CMPXCHG_LOCAL
+        local_irq_restore(flags);
+#endif
         stat(s, FREE_SLAB);
         discard_slab(s, page);
         return;
@@ -1909,21 +2063,54 @@ static __always_inline void slab_free(struct kmem_cache *s,
 {
         void **object = (void *)x;
         struct kmem_cache_cpu *c;
+#ifdef CONFIG_CMPXCHG_LOCAL
+        unsigned long tid;
+#else
         unsigned long flags;
+#endif
 
         slab_free_hook(s, x);
 
+#ifndef CONFIG_CMPXCHG_LOCAL
         local_irq_save(flags);
+#endif
+
+redo:
+        /*
+         * Determine the currently cpus per cpu slab.
+         * The cpu may change afterward. However that does not matter since
+         * data is retrieved via this pointer. If we are on the same cpu
+         * during the cmpxchg then the free will succedd.
+         */
         c = __this_cpu_ptr(s->cpu_slab);
 
+#ifdef CONFIG_CMPXCHG_LOCAL
+        tid = c->tid;
+        barrier();
+#endif
+
         if (likely(page == c->page && c->node != NUMA_NO_NODE)) {
                 set_freepointer(s, object, c->freelist);
+
+#ifdef CONFIG_CMPXCHG_LOCAL
+                if (unlikely(!this_cpu_cmpxchg_double(
+                                s->cpu_slab->freelist, s->cpu_slab->tid,
+                                c->freelist, tid,
+                                object, next_tid(tid)))) {
+
+                        note_cmpxchg_failure("slab_free", s, tid);
+                        goto redo;
+                }
+#else
                 c->freelist = object;
+#endif
                 stat(s, FREE_FASTPATH);
         } else
                 __slab_free(s, page, x, addr);
 
+#ifndef CONFIG_CMPXCHG_LOCAL
         local_irq_restore(flags);
+#endif
 }
 
 void kmem_cache_free(struct kmem_cache *s, void *x)
@@ -2115,9 +2302,23 @@ static inline int alloc_kmem_cache_cpus(struct kmem_cache *s)
         BUILD_BUG_ON(PERCPU_DYNAMIC_EARLY_SIZE <
                         SLUB_PAGE_SHIFT * sizeof(struct kmem_cache_cpu));
 
+#ifdef CONFIG_CMPXCHG_LOCAL
+        /*
+         * Must align to double word boundary for the double cmpxchg instructions
+         * to work.
+         */
+        s->cpu_slab = __alloc_percpu(sizeof(struct kmem_cache_cpu), 2 * sizeof(void *));
+#else
+        /* Regular alignment is sufficient */
         s->cpu_slab = alloc_percpu(struct kmem_cache_cpu);
+#endif
+
+        if (!s->cpu_slab)
+                return 0;
+
+        init_kmem_cache_cpus(s);
 
-        return s->cpu_slab != NULL;
+        return 1;
 }
 
 static struct kmem_cache *kmem_cache_node;