failslab: add ability to filter slab caches

This patch allow to inject faults only for specific slabs.
In order to preserve default behavior cache filter is off by
default (all caches are faulty).

One may define specific set of slabs like this:
# mark skbuff_head_cache as faulty
echo 1 > /sys/kernel/slab/skbuff_head_cache/failslab
# Turn on cache filter (off by default)
echo 1 > /sys/kernel/debug/failslab/cache-filter
# Turn on fault injection
echo 1 > /sys/kernel/debug/failslab/times
echo 1 > /sys/kernel/debug/failslab/probability

Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Akinobu Mita <akinobu.mita@gmail.com>
Acked-by: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>

Documentation/vm/slub.txt

@@ -41,6 +41,7 @@ Possible debug options are
 	P		Poisoning (object and padding)
 	U		User tracking (free and alloc)
 	T		Trace (please only use on single slabs)
+	A		Toggle failslab filter mark for the cache
 	O		Switch debugging off for caches that would have
 			caused higher minimum slab orders
 	-		Switch all debugging off (useful if the kernel is

include/linux/fault-inject.h

@@ -82,9 +82,10 @@ static inline void cleanup_fault_attr_dentries(struct fault_attr *attr)
 #endif /* CONFIG_FAULT_INJECTION */
 
 #ifdef CONFIG_FAILSLAB
-extern bool should_failslab(size_t size, gfp_t gfpflags);
+extern bool should_failslab(size_t size, gfp_t gfpflags, unsigned long flags);
 #else
-static inline bool should_failslab(size_t size, gfp_t gfpflags)
+static inline bool should_failslab(size_t size, gfp_t gfpflags,
+				unsigned long flags)
 {
 	return false;
 }

include/linux/slab.h

@@ -70,6 +70,11 @@
 #else
 # define SLAB_NOTRACK		0x00000000UL
 #endif
+#ifdef CONFIG_FAILSLAB
+# define SLAB_FAILSLAB		0x02000000UL	/* Fault injection mark */
+#else
+# define SLAB_FAILSLAB		0x00000000UL
+#endif
 
 /* The following flags affect the page allocator grouping pages by mobility */
 #define SLAB_RECLAIM_ACCOUNT	0x00020000UL		/* Objects are reclaimable */

mm/failslab.c

@@ -1,25 +1,32 @@
 #include <linux/fault-inject.h>
 #include <linux/gfp.h>
+#include <linux/slab.h>
 
 static struct {
 	struct fault_attr attr;
 	u32 ignore_gfp_wait;
+	int cache_filter;
 #ifdef CONFIG_FAULT_INJECTION_DEBUG_FS
 	struct dentry *ignore_gfp_wait_file;
+	struct dentry *cache_filter_file;
 #endif
 } failslab = {
 	.attr = FAULT_ATTR_INITIALIZER,
 	.ignore_gfp_wait = 1,
+	.cache_filter = 0,
 };
 
-bool should_failslab(size_t size, gfp_t gfpflags)
+bool should_failslab(size_t size, gfp_t gfpflags, unsigned long cache_flags)
 {
 	if (gfpflags & __GFP_NOFAIL)
 		return false;
 
         if (failslab.ignore_gfp_wait && (gfpflags & __GFP_WAIT))
 		return false;
 
+	if (failslab.cache_filter && !(cache_flags & SLAB_FAILSLAB))
+		return false;
+
 	return should_fail(&failslab.attr, size);
 }
 
@@ -30,7 +37,6 @@ static int __init setup_failslab(char *str)
 __setup("failslab=", setup_failslab);
 
 #ifdef CONFIG_FAULT_INJECTION_DEBUG_FS
-
 static int __init failslab_debugfs_init(void)
 {
 	mode_t mode = S_IFREG | S_IRUSR | S_IWUSR;
@@ -46,8 +52,14 @@ static int __init failslab_debugfs_init(void)
 		debugfs_create_bool("ignore-gfp-wait", mode, dir,
 				      &failslab.ignore_gfp_wait);
 
-	if (!failslab.ignore_gfp_wait_file) {
+	failslab.cache_filter_file =
+		debugfs_create_bool("cache-filter", mode, dir,
+				      &failslab.cache_filter);
+
+	if (!failslab.ignore_gfp_wait_file ||
+	    !failslab.cache_filter_file) {
 		err = -ENOMEM;
+		debugfs_remove(failslab.cache_filter_file);
 		debugfs_remove(failslab.ignore_gfp_wait_file);
 		cleanup_fault_attr_dentries(&failslab.attr);
 	}

mm/slab.c

@@ -3101,7 +3101,7 @@ static bool slab_should_failslab(struct kmem_cache *cachep, gfp_t flags)
 	if (cachep == &cache_cache)
 		return false;
 
-	return should_failslab(obj_size(cachep), flags);
+	return should_failslab(obj_size(cachep), flags, cachep->flags);
 }
 
 static inline void *____cache_alloc(struct kmem_cache *cachep, gfp_t flags)

mm/slub.c

@@ -151,7 +151,8 @@
  * Set of flags that will prevent slab merging
  */
 #define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
-		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE)
+		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE | \
+		SLAB_FAILSLAB)
 
 #define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
 		SLAB_CACHE_DMA | SLAB_NOTRACK)
@@ -1020,6 +1021,9 @@ static int __init setup_slub_debug(char *str)
 		case 't':
 			slub_debug |= SLAB_TRACE;
 			break;
+		case 'a':
+			slub_debug |= SLAB_FAILSLAB;
+			break;
 		default:
 			printk(KERN_ERR "slub_debug option '%c' "
 				"unknown. skipped\n", *str);
@@ -1718,7 +1722,7 @@ static __always_inline void *slab_alloc(struct kmem_cache *s,
 	lockdep_trace_alloc(gfpflags);
 	might_sleep_if(gfpflags & __GFP_WAIT);
 
-	if (should_failslab(s->objsize, gfpflags))
+	if (should_failslab(s->objsize, gfpflags, s->flags))
 		return NULL;
 
 	local_irq_save(flags);
@@ -4171,6 +4175,23 @@ static ssize_t trace_store(struct kmem_cache *s, const char *buf,
 }
 SLAB_ATTR(trace);
 
+#ifdef CONFIG_FAILSLAB
+static ssize_t failslab_show(struct kmem_cache *s, char *buf)
+{
+	return sprintf(buf, "%d\n", !!(s->flags & SLAB_FAILSLAB));
+}
+
+static ssize_t failslab_store(struct kmem_cache *s, const char *buf,
+							size_t length)
+{
+	s->flags &= ~SLAB_FAILSLAB;
+	if (buf[0] == '1')
+		s->flags |= SLAB_FAILSLAB;
+	return length;
+}
+SLAB_ATTR(failslab);
+#endif
+
 static ssize_t reclaim_account_show(struct kmem_cache *s, char *buf)
 {
 	return sprintf(buf, "%d\n", !!(s->flags & SLAB_RECLAIM_ACCOUNT));
@@ -4467,6 +4488,10 @@ static struct attribute *slab_attrs[] = {
 	&deactivate_remote_frees_attr.attr,
 	&order_fallback_attr.attr,
 #endif
+#ifdef CONFIG_FAILSLAB
+	&failslab_attr.attr,
+#endif
+
 	NULL
 };