[PATCH] NUMA slab locking fixes: irq disabling from cahep->spinlock to l3 lock
authorRavikiran G Thirumalai <kiran@scalex86.org>
Sun, 5 Feb 2006 07:27:58 +0000 (23:27 -0800)
committerLinus Torvalds <torvalds@g5.osdl.org>
Sun, 5 Feb 2006 19:06:53 +0000 (11:06 -0800)
Earlier, we had to disable on chip interrupts while taking the
cachep->spinlock because, at cache_grow, on every addition of a slab to a slab
cache, we incremented colour_next which was protected by the cachep->spinlock,
and cache_grow could occur at interrupt context.  Since, now we protect the
per-node colour_next with the node's list_lock, we do not need to disable on
chip interrupts while taking the per-cache spinlock, but we just need to
disable interrupts when taking the per-node kmem_list3 list_lock.

Signed-off-by: Alok N Kataria <alokk@calsoftinc.com>
Signed-off-by: Ravikiran Thirumalai <kiran@scalex86.org>
Signed-off-by: Shai Fultheim <shai@scalex86.org>
Cc: Christoph Lameter <christoph@lameter.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
mm/slab.c

index 2317096166dde3c632d12a79c7d0772c50f603d5..d3f68543f9f4e3fcb403192357d8d18056b75b03 100644 (file)
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -987,7 +987,7 @@ static int __devinit cpuup_callback(struct notifier_block *nfb,
                        cpumask_t mask;
 
                        mask = node_to_cpumask(node);
-                       spin_lock_irq(&cachep->spinlock);
+                       spin_lock(&cachep->spinlock);
                        /* cpu is dead; no one can alloc from it. */
                        nc = cachep->array[cpu];
                        cachep->array[cpu] = NULL;
@@ -996,7 +996,7 @@ static int __devinit cpuup_callback(struct notifier_block *nfb,
                        if (!l3)
                                goto unlock_cache;
 
-                       spin_lock(&l3->list_lock);
+                       spin_lock_irq(&l3->list_lock);
 
                        /* Free limit for this kmem_list3 */
                        l3->free_limit -= cachep->batchcount;
@@ -1004,7 +1004,7 @@ static int __devinit cpuup_callback(struct notifier_block *nfb,
                                free_block(cachep, nc->entry, nc->avail, node);
 
                        if (!cpus_empty(mask)) {
-                               spin_unlock(&l3->list_lock);
+                               spin_unlock_irq(&l3->list_lock);
                                goto unlock_cache;
                        }
 
@@ -1023,13 +1023,13 @@ static int __devinit cpuup_callback(struct notifier_block *nfb,
                        /* free slabs belonging to this node */
                        if (__node_shrink(cachep, node)) {
                                cachep->nodelists[node] = NULL;
-                               spin_unlock(&l3->list_lock);
+                               spin_unlock_irq(&l3->list_lock);
                                kfree(l3);
                        } else {
-                               spin_unlock(&l3->list_lock);
+                               spin_unlock_irq(&l3->list_lock);
                        }
                      unlock_cache:
-                       spin_unlock_irq(&cachep->spinlock);
+                       spin_unlock(&cachep->spinlock);
                        kfree(nc);
                }
                mutex_unlock(&cache_chain_mutex);
@@ -2011,18 +2011,18 @@ static void drain_cpu_caches(struct kmem_cache *cachep)
 
        smp_call_function_all_cpus(do_drain, cachep);
        check_irq_on();
-       spin_lock_irq(&cachep->spinlock);
+       spin_lock(&cachep->spinlock);
        for_each_online_node(node) {
                l3 = cachep->nodelists[node];
                if (l3) {
-                       spin_lock(&l3->list_lock);
+                       spin_lock_irq(&l3->list_lock);
                        drain_array_locked(cachep, l3->shared, 1, node);
-                       spin_unlock(&l3->list_lock);
+                       spin_unlock_irq(&l3->list_lock);
                        if (l3->alien)
                                drain_alien_cache(cachep, l3);
                }
        }
-       spin_unlock_irq(&cachep->spinlock);
+       spin_unlock(&cachep->spinlock);
 }
 
 static int __node_shrink(struct kmem_cache *cachep, int node)
@@ -2338,7 +2338,6 @@ static int cache_grow(struct kmem_cache *cachep, gfp_t flags, int nodeid)
 
        offset *= cachep->colour_off;
 
-       check_irq_off();
        if (local_flags & __GFP_WAIT)
                local_irq_enable();
 
@@ -2725,6 +2724,7 @@ static void *__cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int node
        BUG_ON(!l3);
 
       retry:
+       check_irq_off();
        spin_lock(&l3->list_lock);
        entry = l3->slabs_partial.next;
        if (entry == &l3->slabs_partial) {
@@ -3304,11 +3304,11 @@ static int do_tune_cpucache(struct kmem_cache *cachep, int limit, int batchcount
        smp_call_function_all_cpus(do_ccupdate_local, (void *)&new);
 
        check_irq_on();
-       spin_lock_irq(&cachep->spinlock);
+       spin_lock(&cachep->spinlock);
        cachep->batchcount = batchcount;
        cachep->limit = limit;
        cachep->shared = shared;
-       spin_unlock_irq(&cachep->spinlock);
+       spin_unlock(&cachep->spinlock);
 
        for_each_online_cpu(i) {
                struct array_cache *ccold = new.new[i];
@@ -3564,8 +3564,7 @@ static int s_show(struct seq_file *m, void *p)
        int node;
        struct kmem_list3 *l3;
 
-       check_irq_on();
-       spin_lock_irq(&cachep->spinlock);
+       spin_lock(&cachep->spinlock);
        active_objs = 0;
        num_slabs = 0;
        for_each_online_node(node) {
@@ -3573,7 +3572,8 @@ static int s_show(struct seq_file *m, void *p)
                if (!l3)
                        continue;
 
-               spin_lock(&l3->list_lock);
+               check_irq_on();
+               spin_lock_irq(&l3->list_lock);
 
                list_for_each(q, &l3->slabs_full) {
                        slabp = list_entry(q, struct slab, list);
@@ -3600,7 +3600,7 @@ static int s_show(struct seq_file *m, void *p)
                free_objects += l3->free_objects;
                shared_avail += l3->shared->avail;
 
-               spin_unlock(&l3->list_lock);
+               spin_unlock_irq(&l3->list_lock);
        }
        num_slabs += active_slabs;
        num_objs = num_slabs * cachep->num;
@@ -3644,7 +3644,7 @@ static int s_show(struct seq_file *m, void *p)
        }
 #endif
        seq_putc(m, '\n');
-       spin_unlock_irq(&cachep->spinlock);
+       spin_unlock(&cachep->spinlock);
        return 0;
 }