Re: [exim-dev] This always worries me

Kezdőlap
Üzenet törlése
Válasz az üzenetre
Szerző: Jeremy Harris
Dátum:  
Címzett: exim-dev
Tárgy: Re: [exim-dev] This always worries me
On 27/09/13 12:44, john ffitch wrote:
> On a Debian machine with X86_64 processors (3.2.0-4-amd64 #1 SMP
> Debian 3.2.46-1 x86_64 GNU/Linux) ....
> I think this has been around for a while but it does indicate something
> wrong.
>
>
> gcc exim_monitor/em_TextPop.c
> ../exim_monitor/em_TextPop.c: In function ‘DoSearch’:
> ../exim_monitor/em_TextPop.c:471:31: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
> gcc exim_monitor/em_main.c
> ../exim_monitor/em_main.c: In function ‘numlock_modifiers’:
> ../exim_monitor/em_main.c:576:5: warning: ‘XKeycodeToKeysym’ is deprecated (declared at /usr/include/X11/Xlib.h:1695) [-Wdeprecated-declarations]
> gcc exim_monitor/em_strip.c
> ../exim_monitor/em_strip.c: In function ‘stripchartAction’:
> ../exim_monitor/em_strip.c:66:11: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
> ../exim_monitor/em_strip.c: In function ‘create_stripchart’:
> ../exim_monitor/em_strip.c:252:3: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]


The whole thing is a horrid mess of casting. Probably this is inherent
in X...

Does the attached patch quieten it down any for you?
-- 
Cheers,
    Jeremy

diff --git a/ideas-notes b/ideas-notes
new file mode 100644
index 0000000..b97243b
--- /dev/null
+++ b/ideas-notes
@@ -0,0 +1,25 @@
+Auto-slab:  interception wrapper on kmalloc, auto-creating
+(by size) a slab labelled by module & size.
+- _RET_IP_ gives caller address [cf. kmem_cache_alloc()]
+- use kallsyms_lookup() on the caller address
+  - gives module name; use as part of slab name
+- use size as part of slab name
+- hash caller address for fast lookup
+- only create slab after (tunable?) a few calls
+  - kmem_cache_create()   creates a slab
+- must intercept both kmalloc() & kfree()
+  - kmem_cache_alloc()/kmem_cache_free() are the replacements we want to call
+  - maybe a duplicate of kmem_cache_alloc() so as to get a better
+    caller-IP logged?
+? number of elements in slab?
+? do we nee a module unload hook to delete the slab?
+  - kmem_cache_destroy()
+  - if we don't, module reloads may be an issue
+  - if we don't, kernel shutdowns may be an issue
+  - must also clear out hashtable entries
+- alloc must support DMA, ATOMIC (etc.) flags
+
+
+Installer: precalculated dependencies for each offered standard
+install set.
+
diff --git a/include/linux/slab.h b/include/linux/slab.h
index 0c62175..4e40091 100644
--- a/include/linux/slab.h
+++ b/include/linux/slab.h
@@ -155,7 +155,7 @@ size_t ksize(const void *);
  *
  * Once we can do anonymous structs (C11 standard) we could put a
  * anonymous struct definition in these allocators so that the
- * separate allocations in the kmem_cache structure of SLAB and
+ * separate allocations in the kmem_cache structure of SLAB, SLEB and
  * SLUB is no longer needed.
  */
 struct kmem_cache {
@@ -197,7 +197,7 @@ struct kmem_cache {
 #endif
 #else
 /*
- * SLUB allocates up to order 2 pages directly and otherwise
+ * SLUB/SLEB allocates up to order 2 pages directly and otherwise
  * passes the request to the page allocator.
  */
 #define KMALLOC_SHIFT_HIGH    (PAGE_SHIFT + 1)
@@ -280,6 +280,8 @@ static __always_inline int kmalloc_index(size_t size)
 #include <linux/slab_def.h>
 #elif defined(CONFIG_SLUB)
 #include <linux/slub_def.h>
+#elif defined(CONFIG_SLEB)
+#include <linux/sleb_def.h>
 #else
 #error "Unknown slab allocator"
 #endif
@@ -463,6 +465,7 @@ static inline void *kmem_cache_alloc_node(struct kmem_cache *cachep,
  * request comes from.
  */
 #if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_SLUB) || \
+    defined(CONFIG_SLEB) || \
     (defined(CONFIG_SLAB) && defined(CONFIG_TRACING)) || \
     (defined(CONFIG_SLOB) && defined(CONFIG_TRACING))
 extern void *__kmalloc_track_caller(size_t, gfp_t, unsigned long);
@@ -483,6 +486,7 @@ extern void *__kmalloc_track_caller(size_t, gfp_t, unsigned long);
  * allocation request comes from.
  */
 #if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_SLUB) || \
+    defined(CONFIG_SLEB) || \
     (defined(CONFIG_SLAB) && defined(CONFIG_TRACING)) || \
     (defined(CONFIG_SLOB) && defined(CONFIG_TRACING))
 extern void *__kmalloc_node_track_caller(size_t, gfp_t, int, unsigned long);
diff --git a/include/linux/sleb_def.h b/include/linux/sleb_def.h
new file mode 100644
index 0000000..a35a28d
--- /dev/null
+++ b/include/linux/sleb_def.h
@@ -0,0 +1,318 @@
+#ifndef _LINUX_SLEB_DEF_H
+#define _LINUX_SLEB_DEF_H
+
+/*
+ * SLEB : A Slab allocator without object queues.
+ *
+ * (C) 2007 SGI, Christoph Lameter
+ */
+#include <linux/types.h>
+#include <linux/gfp.h>
+#include <linux/bug.h>
+#include <linux/workqueue.h>
+#include <linux/kobject.h>
+#include <linux/module.h>
+
+#include <linux/kmemleak.h>
+
+enum stat_item {
+    ALLOC_FASTPATH,        /* Allocation from cpu slab */
+    ALLOC_SLOWPATH,        /* Allocation by getting a new cpu slab */
+    FREE_FASTPATH,        /* Free to cpu slab */
+    FREE_SLOWPATH,        /* Freeing not to cpu slab */
+    FREE_FROZEN,        /* Freeing to frozen slab */
+    FREE_ADD_PARTIAL,    /* Freeing moves slab to partial list */
+    FREE_REMOVE_PARTIAL,    /* Freeing removes last object */
+    ALLOC_FROM_PARTIAL,    /* Cpu slab acquired from node partial list */
+    ALLOC_SLAB,        /* Cpu slab acquired from page allocator */
+    ALLOC_REFILL,        /* Refill cpu slab from slab freelist */
+    ALLOC_NODE_MISMATCH,    /* Switching cpu slab */
+    FREE_SLAB,        /* Slab freed to the page allocator */
+    CPUSLAB_FLUSH,        /* Abandoning of the cpu slab */
+    DEACTIVATE_FULL,    /* Cpu slab was full when deactivated */
+    DEACTIVATE_EMPTY,    /* Cpu slab was empty when deactivated */
+    DEACTIVATE_TO_HEAD,    /* Cpu slab was moved to the head of partials */
+    DEACTIVATE_TO_TAIL,    /* Cpu slab was moved to the tail of partials */
+    DEACTIVATE_REMOTE_FREES,/* Slab contained remotely freed objects */
+    DEACTIVATE_BYPASS,    /* Implicit deactivation */
+    ORDER_FALLBACK,        /* Number of times fallback was necessary */
+    CMPXCHG_DOUBLE_CPU_FAIL,/* Failure of this_cpu_cmpxchg_double */
+    CMPXCHG_DOUBLE_FAIL,    /* Number of times that cmpxchg double did not match */
+    CPU_PARTIAL_ALLOC,    /* Used cpu partial on alloc */
+    CPU_PARTIAL_FREE,    /* Refill cpu partial on free */
+    CPU_PARTIAL_NODE,    /* Refill cpu partial from node partial */
+    CPU_PARTIAL_DRAIN,    /* Drain cpu partial to node partial */
+    NR_SLEB_STAT_ITEMS };
+
+struct kmem_cache_cpu {
+    void **freelist;    /* Pointer to next available object */
+    unsigned long tid;    /* Globally unique transaction id */
+    struct page *page;    /* The slab from which we are allocating */
+    struct page *partial;    /* Partially allocated frozen slabs */
+#ifdef CONFIG_SLEB_STATS
+    unsigned stat[NR_SLEB_STAT_ITEMS];
+#endif
+};
+
+/*
+ * Word size structure that can be atomically updated or read and that
+ * contains both the order and the number of objects that a slab of the
+ * given order would contain.
+ */
+struct kmem_cache_order_objects {
+    unsigned long x;
+};
+
+/*
+ * Slab cache management.
+ */
+struct kmem_cache {
+    struct kmem_cache_cpu __percpu *cpu_slab;
+    /* Used for retriving partial slabs etc */
+    unsigned long flags;
+    unsigned long min_partial;
+    int size;        /* The size of an object including meta data */
+    int object_size;    /* The size of an object without meta data */
+    int offset;        /* Free pointer offset. */
+    int cpu_partial;    /* Number of per cpu partial objects to keep around */
+    struct kmem_cache_order_objects oo;
+
+    /* Allocation and freeing of slabs */
+    struct kmem_cache_order_objects max;
+    struct kmem_cache_order_objects min;
+    gfp_t allocflags;    /* gfp flags to use on each alloc */
+    int refcount;        /* Refcount for slab cache destroy */
+    void (*ctor)(void *);
+    int inuse;        /* Offset to metadata */
+    int align;        /* Alignment */
+    int reserved;        /* Reserved bytes at the end of slabs */
+    const char *name;    /* Name (only for display!) */
+    struct list_head list;    /* List of slab caches */
+#ifdef CONFIG_SYSFS
+    struct kobject kobj;    /* For sysfs */
+#endif
+#ifdef CONFIG_MEMCG_KMEM
+    struct memcg_cache_params *memcg_params;
+    int max_attr_size; /* for propagation, maximum size of a stored attr */
+#endif
+
+#ifdef CONFIG_NUMA
+    /*
+     * Defragmentation by allocating from a remote node.
+     */
+    int remote_node_defrag_ratio;
+#endif
+    struct kmem_cache_node *node[MAX_NUMNODES];
+};
+
+void *kmem_cache_alloc(struct kmem_cache *, gfp_t);
+void *__kmalloc(size_t size, gfp_t flags);
+
+static __always_inline void *
+kmalloc_order(size_t size, gfp_t flags, unsigned int order)
+{
+    void *ret;
+
+    flags |= (__GFP_COMP | __GFP_KMEMCG);
+    ret = (void *) __get_free_pages(flags, order);
+    kmemleak_alloc(ret, size, 1, flags);
+    return ret;
+}
+
+/**
+ * Calling this on allocated memory will check that the memory
+ * is expected to be in use, and print warnings if not.
+ */
+#ifdef CONFIG_SLEB_DEBUG
+extern bool verify_mem_not_deleted(const void *x);
+#else
+static inline bool verify_mem_not_deleted(const void *x)
+{
+    return true;
+}
+#endif
+
+#ifdef CONFIG_TRACING
+extern void *
+kmem_cache_alloc_trace(struct kmem_cache *s, gfp_t gfpflags, size_t size);
+extern void *kmalloc_order_trace(size_t size, gfp_t flags, unsigned int order);
+#else
+static __always_inline void *
+kmem_cache_alloc_trace(struct kmem_cache *s, gfp_t gfpflags, size_t size)
+{
+    return kmem_cache_alloc(s, gfpflags);
+}
+
+static __always_inline void *
+kmalloc_order_trace(size_t size, gfp_t flags, unsigned int order)
+{
+    return kmalloc_order(size, flags, order);
+}
+#endif
+
+static __always_inline void *kmalloc_large(size_t size, gfp_t flags)
+{
+    unsigned int order = get_order(size);
+    return kmalloc_order_trace(size, flags, order);
+}
+
+
+
+struct sleb_caller_entry {
+    unsigned long    caller;
+    struct sleb_module_entry *mp;
+};
+
+/*XXX do we need an init for this?  zero out caller fields. */
+#define SLEB_CALLER_HASH_SIZE 1024
+static struct sleb_caller_entry sleb_caller_hash[SLEB_CALLER_HASH_SIZE];
+#define SLEB_CALLER_HASHIDX(caller) ((caller >> 5) & (SLEB_CALLER_HASH_SIZE-1))
+
+struct sleb_module_entry {
+    size_t        size;
+    char        modname[MODULE_NAME_LEN];
+    unsigned    calls;
+    struct kmem_cache *slab;
+};
+
+#define SLEB_MODULE_HASH_SIZE 512
+static struct sleb_module_entry sleb_module_hash[SLEB_MODULE_HASH_SIZE];
+static __always_inline unsigned sleb_module_hashidx(
+                                        const char *modname, size_t size)
+{
+    int i = SLEB_MODULE_HASH_SIZE;
+    unsigned ret = (unsigned)(size % SLEB_MODULE_HASH_SIZE);
+    for( i = SLEB_MODULE_HASH_SIZE; i>=8; i-=8 )    /*XXX could we do faster */
+        ret = (ret<<8) | *modname++;                /*    when aligned?      */
+    return (ret<<8) |  (*modname & ((1<<i)-1));
+}
+
+
+static __always_inline struct kmem_cache *
+frequent_caller(size_t size, unsigned long caller)
+{
+    struct sleb_caller_entry *ep =
+         &sleb_caller_hash[ SLEB_CALLER_HASHIDX(caller) ];
+    char *modname;
+    const char *symname;
+    struct sleb_module_entry *mp;
+    char slabname[MODULE_NAME_LEN];
+
+    if (!ep->caller)
+    {                    /* new caller */
+        ep->caller = caller;
+        symname = module_address_lookup(caller,
+                        NULL, NULL, &modname, NULL);
+        if (!symname) {
+            printk("%s: caller not in module", __FUNCTION__);
+            return NULL;    /* Caller not in a module */
+        }
+        printk("%s: new caller, %s", __FUNCTION__, modname);
+
+        mp = &sleb_module_hash[ sleb_module_hashidx(modname, size) ];
+        if ( !mp->size ) {    /* no existing module entry */
+            printk("%s: new module, %s", __FUNCTION__, modname);
+            ep->mp = mp;
+            mp->size = size;
+            strncpy(mp->modname, modname, MODULE_NAME_LEN-1);
+            mp->calls = 1;
+            return NULL;    /* Not a high-volume caller */
+        }
+
+        if ( size != mp->size || strcmp(mp->modname, modname) != 0 ) {
+                            /* module hash collision */
+            printk("%s: %s collision", __FUNCTION__,
+                    size!=mp->size ? "size":"modname");
+            return NULL;    /*XXX no rehash yet.  How about a counter? */
+        }
+        ep->mp = mp;        /* new caller for existing module */
+    }
+    else if (ep->caller != caller) {    /* caller hash collision */
+        printk("%s: caller collision", __FUNCTION__);
+        return NULL;    /*XXX no rehash yet */
+    } else {                /* existing caller */
+        symname = module_address_lookup(caller,
+                        NULL, NULL, &modname, NULL);
+        if (!symname)
+            return NULL;    /* Caller not in a module */
+        mp = ep->mp;
+        /*assert(mp->size == size);*/
+        /*assert(strcmp(mp->modname, modname) == 0);*/
+    }
+
+    if (mp->slab)
+        return mp->slab; /* slab ready for use */
+
+    if (mp->calls < 10) {
+        printk("%s: low-vol caller", __FUNCTION__);
+        mp->calls++;
+        return NULL;    /* Not a high-volume caller */
+    }
+
+    /* create slab */
+    snprintf(slabname, sizeof(slabname), "%s_%lu",
+                mp->modname, (long unsigned)size);
+    return mp->slab = kmem_cache_create(slabname, size, 0, 0, NULL);
+}
+
+static __always_inline void *kmalloc(size_t size, gfp_t flags)
+{
+    if (__builtin_constant_p(size)) {
+        if (size > KMALLOC_MAX_CACHE_SIZE)
+            return kmalloc_large(size, flags);
+
+        if (!(flags & GFP_DMA)) {
+            struct kmem_cache *s;
+            int index;
+
+            s = frequent_caller(size, _THIS_IP_);
+            if (s)
+                return kmem_cache_alloc_trace(s,
+                        flags, size);
+
+            index = kmalloc_index(size);
+            if (!index)
+                return ZERO_SIZE_PTR;
+
+            return kmem_cache_alloc_trace(kmalloc_caches[index],
+                    flags, size);
+        }
+    }
+    return __kmalloc(size, flags);
+}
+
+#ifdef CONFIG_NUMA
+void *__kmalloc_node(size_t size, gfp_t flags, int node);
+void *kmem_cache_alloc_node(struct kmem_cache *, gfp_t flags, int node);
+
+#ifdef CONFIG_TRACING
+extern void *kmem_cache_alloc_node_trace(struct kmem_cache *s,
+                       gfp_t gfpflags,
+                       int node, size_t size);
+#else
+static __always_inline void *
+kmem_cache_alloc_node_trace(struct kmem_cache *s,
+                  gfp_t gfpflags,
+                  int node, size_t size)
+{
+    return kmem_cache_alloc_node(s, gfpflags, node);
+}
+#endif
+
+static __always_inline void *kmalloc_node(size_t size, gfp_t flags, int node)
+{
+    if (__builtin_constant_p(size) &&
+        size <= KMALLOC_MAX_CACHE_SIZE && !(flags & GFP_DMA)) {
+        int index = kmalloc_index(size);
+
+        if (!index)
+            return ZERO_SIZE_PTR;
+
+        return kmem_cache_alloc_node_trace(kmalloc_caches[index],
+                   flags, node, size);
+    }
+    return __kmalloc_node(size, flags, node);
+}
+#endif
+
+#endif /* _LINUX_SLEB_DEF_H */
diff --git a/init/Kconfig b/init/Kconfig
index a2f4a75..e78a37e 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -923,7 +923,7 @@ config MEMCG_SWAP_ENABLED
 config MEMCG_KMEM
     bool "Memory Resource Controller Kernel Memory accounting"
     depends on MEMCG
-    depends on SLUB || SLAB
+    depends on SLUB || SLAB || SLEB
     help
       The Kernel Memory extension for Memory Resource Controller can limit
       the amount of memory used by kernel objects in the system. Those are
@@ -1498,6 +1498,16 @@ config SLUB_DEBUG
       SLUB sysfs support. /sys/slab will not exist and there will be
       no support for cache validation etc.


+config SLEB_DEBUG
+    default y
+    bool "Enable SLEB debugging support" if EXPERT
+    depends on SLEB && SYSFS
+    help
+      SLEB has extensive debug support features. Disabling these can
+      result in significant savings in code size. This also disables
+      SLEB sysfs support. /sys/slab will not exist and there will be
+      no support for cache validation etc.
+
 config COMPAT_BRK
     bool "Disable heap randomization"
     default y
@@ -1533,6 +1543,12 @@ config SLUB
        and has enhanced diagnostics. SLUB is the default choice for
        a slab allocator.


+config SLEB
+    bool "SLEB (Extended SLUB Allocator)"
+    help
+       SLEB is a slab allocator that behaves like SLUB and addtionally
+       auto-creates slabs for callers of kmalloc.
+
 config SLOB
     depends on EXPERT
     bool "SLOB (Simple Allocator)"
@@ -1589,7 +1605,7 @@ config HAVE_GENERIC_DMA_COHERENT
 config SLABINFO
     bool
     depends on PROC_FS
-    depends on SLAB || SLUB_DEBUG
+    depends on SLAB || SLUB_DEBUG || SLEB_DEBUG
     default y


 config RT_MUTEXES
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 566cf2b..50a6e52 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -449,6 +449,32 @@ config SLUB_STATS
       out which slabs are relevant to a particular load.
       Try running: slabinfo -DA


+config SLEB_DEBUG_ON
+    bool "SLEB debugging on by default"
+    depends on SLEB && SLEB_DEBUG && !KMEMCHECK
+    default n
+    help
+      Boot with debugging on by default. SLEB boots by default with
+      the runtime debug capabilities switched off. Enabling this is
+      equivalent to specifying the "slub_debug" parameter on boot.
+      There is no support for more fine grained debug control like
+      possible with slub_debug=xxx. SLEB debugging may be switched
+      off in a kernel built with CONFIG_SLEB_DEBUG_ON by specifying
+      "slub_debug=-".
+
+config SLEB_STATS
+    default n
+    bool "Enable SLEB performance statistics"
+    depends on SLEB && SYSFS
+    help
+      SLEB statistics are useful to debug SLEBs allocation behavior in
+      order find ways to optimize the allocator. This should never be
+      enabled for production use since keeping statistics slows down
+      the allocator by a few percentage points. The slabinfo command
+      supports the determination of the most active slabs to figure
+      out which slabs are relevant to a particular load.
+      Try running: slabinfo -DA
+
 config HAVE_DEBUG_KMEMLEAK
     bool


@@ -469,8 +495,8 @@ config DEBUG_KMEMLEAK
       allocations. See Documentation/kmemleak.txt for more
       details.


-      Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
-      of finding leaks due to the slab objects poisoning.
+      Enabling DEBUG_SLAB, SLUB_DEBUG or SLEB_DEBUG may increase
+      the chances of finding leaks due to the slab objects poisoning.


       In order to access the kmemleak file, debugfs needs to be
       mounted (usually at /sys/kernel/debug).
@@ -1223,7 +1249,7 @@ config FAULT_INJECTION
 config FAILSLAB
     bool "Fault-injection capability for kmalloc"
     depends on FAULT_INJECTION
-    depends on SLAB || SLUB
+    depends on SLAB || SLUB || SLEB
     help
       Provide fault-injection capability for kmalloc.


diff --git a/lib/Kconfig.kmemcheck b/lib/Kconfig.kmemcheck
index 846e039..627b003 100644
--- a/lib/Kconfig.kmemcheck
+++ b/lib/Kconfig.kmemcheck
@@ -7,7 +7,7 @@ menuconfig KMEMCHECK
     bool "kmemcheck: trap use of uninitialized memory"
     depends on DEBUG_KERNEL
     depends on !X86_USE_3DNOW
-    depends on SLUB || SLAB
+    depends on SLUB || SLAB || SLEB
     depends on !CC_OPTIMIZE_FOR_SIZE
     depends on !FUNCTION_TRACER
     select FRAME_POINTER
diff --git a/mm/Makefile b/mm/Makefile
index f008033..4012697 100644
--- a/mm/Makefile
+++ b/mm/Makefile
@@ -43,6 +43,7 @@ obj-$(CONFIG_MMU_NOTIFIER) += mmu_notifier.o
 obj-$(CONFIG_KSM) += ksm.o
 obj-$(CONFIG_PAGE_POISONING) += debug-pagealloc.o
 obj-$(CONFIG_SLAB) += slab.o
+obj-$(CONFIG_SLEB) += sleb.o
 obj-$(CONFIG_SLUB) += slub.o
 obj-$(CONFIG_KMEMCHECK) += kmemcheck.o
 obj-$(CONFIG_FAILSLAB) += failslab.o
diff --git a/mm/dmapool.c b/mm/dmapool.c
index c69781e..126aeac 100644
--- a/mm/dmapool.c
+++ b/mm/dmapool.c
@@ -38,7 +38,7 @@
 #include <linux/types.h>
 #include <linux/wait.h>


-#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_SLUB_DEBUG_ON)
+#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_SLUB_DEBUG_ON) || defined(CONFIG_SLEB_DEBUG_ON)
#define DMAPOOL_DEBUG 1
#endif

diff --git a/mm/slab.h b/mm/slab.h
index f96b49e..a2db182 100644
--- a/mm/slab.h
+++ b/mm/slab.h
@@ -53,7 +53,7 @@ extern void create_boot_cache(struct kmem_cache *, const char *name,
             size_t size, unsigned long flags);


 struct mem_cgroup;
-#ifdef CONFIG_SLUB
+#if defined(CONFIG_SLUB) || defined(CONFIG_SLEB)
 struct kmem_cache *
 __kmem_cache_alias(struct mem_cgroup *memcg, const char *name, size_t size,
            size_t align, unsigned long flags, void (*ctor)(void *));
@@ -71,7 +71,7 @@ __kmem_cache_alias(struct mem_cgroup *memcg, const char *name, size_t size,


 #if defined(CONFIG_DEBUG_SLAB)
 #define SLAB_DEBUG_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
-#elif defined(CONFIG_SLUB_DEBUG)
+#elif defined(CONFIG_SLUB_DEBUG) || defined(CONFIG_SLEB_DEBUG)
 #define SLAB_DEBUG_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
               SLAB_TRACE | SLAB_DEBUG_FREE)
 #else
@@ -81,7 +81,7 @@ __kmem_cache_alias(struct mem_cgroup *memcg, const char *name, size_t size,
 #if defined(CONFIG_SLAB)
 #define SLAB_CACHE_FLAGS (SLAB_MEM_SPREAD | SLAB_NOLEAKTRACE | \
               SLAB_RECLAIM_ACCOUNT | SLAB_TEMPORARY | SLAB_NOTRACK)
-#elif defined(CONFIG_SLUB)
+#elif defined(CONFIG_SLUB) || defined(CONFIG_SLEB)
 #define SLAB_CACHE_FLAGS (SLAB_NOLEAKTRACE | SLAB_RECLAIM_ACCOUNT | \
               SLAB_TEMPORARY | SLAB_NOTRACK)
 #else
@@ -270,4 +270,14 @@ struct kmem_cache_node {
 #endif
 #endif


+#ifdef CONFIG_SLEB
+    unsigned long nr_partial;
+    struct list_head partial;
+#ifdef CONFIG_SLEB_DEBUG
+    atomic_long_t nr_slabs;
+    atomic_long_t total_objects;
+    struct list_head full;
+#endif
+#endif
+
 };
diff --git a/mm/sleb.c b/mm/sleb.c
new file mode 100644
index 0000000..10f578d
--- /dev/null
+++ b/mm/sleb.c
@@ -0,0 +1,5309 @@
+/*
+ * SLEB: A SLUB variant that creates named slabs for users of kmalloc.
+ * SLUB is a slab allocator that limits cache line use instead of queuing
+ * objects in per cpu and per node lists.
+ *
+ * The allocator synchronizes using per slab locks or atomic operatios
+ * and only uses a centralized lock to manage a pool of partial slabs.
+ *
+ * (C) 2007 SGI, Christoph Lameter
+ * (C) 2011 Linux Foundation, Christoph Lameter
+ * (C) 2013 Red Hat, Jeremy Harris
+ */
+
+#include <linux/mm.h>
+#include <linux/swap.h> /* struct reclaim_state */
+#include <linux/module.h>
+#include <linux/bit_spinlock.h>
+#include <linux/interrupt.h>
+#include <linux/bitops.h>
+#include <linux/slab.h>
+#include "slab.h"
+#include <linux/proc_fs.h>
+#include <linux/notifier.h>
+#include <linux/seq_file.h>
+#include <linux/kmemcheck.h>
+#include <linux/cpu.h>
+#include <linux/cpuset.h>
+#include <linux/mempolicy.h>
+#include <linux/ctype.h>
+#include <linux/debugobjects.h>
+#include <linux/kallsyms.h>
+#include <linux/memory.h>
+#include <linux/math64.h>
+#include <linux/fault-inject.h>
+#include <linux/stacktrace.h>
+#include <linux/prefetch.h>
+#include <linux/memcontrol.h>
+
+#include <trace/events/kmem.h>
+
+#include "internal.h"
+
+/*
+ * Lock order:
+ *   1. slab_mutex (Global Mutex)
+ *   2. node->list_lock
+ *   3. slab_lock(page) (Only on some arches and for debugging)
+ *
+ *   slab_mutex
+ *
+ *   The role of the slab_mutex is to protect the list of all the slabs
+ *   and to synchronize major metadata changes to slab cache structures.
+ *
+ *   The slab_lock is only used for debugging and on arches that do not
+ *   have the ability to do a cmpxchg_double. It only protects the second
+ *   double word in the page struct. Meaning
+ *    A. page->freelist    -> List of object free in a page
+ *    B. page->counters    -> Counters of objects
+ *    C. page->frozen        -> frozen state
+ *
+ *   If a slab is frozen then it is exempt from list management. It is not
+ *   on any list. The processor that froze the slab is the one who can
+ *   perform list operations on the page. Other processors may put objects
+ *   onto the freelist but the processor that froze the slab is the only
+ *   one that can retrieve the objects from the page's freelist.
+ *
+ *   The list_lock protects the partial and full list on each node and
+ *   the partial slab counter. If taken then no new slabs may be added or
+ *   removed from the lists nor make the number of partial slabs be modified.
+ *   (Note that the total number of slabs is an atomic value that may be
+ *   modified without taking the list lock).
+ *
+ *   The list_lock is a centralized lock and thus we avoid taking it as
+ *   much as possible. As long as SLUB does not have to handle partial
+ *   slabs, operations can continue without any centralized lock. F.e.
+ *   allocating a long series of objects that fill up slabs does not require
+ *   the list lock.
+ *   Interrupts are disabled during allocation and deallocation in order to
+ *   make the slab allocator safe to use in the context of an irq. In addition
+ *   interrupts are disabled to ensure that the processor does not change
+ *   while handling per_cpu slabs, due to kernel preemption.
+ *
+ * SLUB assigns one slab for allocation to each processor.
+ * Allocations only occur from these slabs called cpu slabs.
+ *
+ * Slabs with free elements are kept on a partial list and during regular
+ * operations no list for full slabs is used. If an object in a full slab is
+ * freed then the slab will show up again on the partial lists.
+ * We track full slabs for debugging purposes though because otherwise we
+ * cannot scan all objects.
+ *
+ * Slabs are freed when they become empty. Teardown and setup is
+ * minimal so we rely on the page allocators per cpu caches for
+ * fast frees and allocs.
+ *
+ * Overloading of page flags that are otherwise used for LRU management.
+ *
+ * PageActive         The slab is frozen and exempt from list processing.
+ *             This means that the slab is dedicated to a purpose
+ *             such as satisfying allocations for a specific
+ *             processor. Objects may be freed in the slab while
+ *             it is frozen but slab_free will then skip the usual
+ *             list operations. It is up to the processor holding
+ *             the slab to integrate the slab into the slab lists
+ *             when the slab is no longer needed.
+ *
+ *             One use of this flag is to mark slabs that are
+ *             used for allocations. Then such a slab becomes a cpu
+ *             slab. The cpu slab may be equipped with an additional
+ *             freelist that allows lockless access to
+ *             free objects in addition to the regular freelist
+ *             that requires the slab lock.
+ *
+ * PageError        Slab requires special handling due to debug
+ *             options set. This moves    slab handling out of
+ *             the fast path and disables lockless freelists.
+ */
+
+static inline int kmem_cache_debug(struct kmem_cache *s)
+{
+#ifdef CONFIG_SLEB_DEBUG
+    return unlikely(s->flags & SLAB_DEBUG_FLAGS);
+#else
+    return 0;
+#endif
+}
+
+/*
+ * Issues still to be resolved:
+ *
+ * - Support PAGE_ALLOC_DEBUG. Should be easy to do.
+ *
+ * - Variable sizing of the per node arrays
+ */
+
+/* Enable to test recovery from slab corruption on boot */
+#undef SLUB_RESILIENCY_TEST
+
+/* Enable to log cmpxchg failures */
+#undef SLUB_DEBUG_CMPXCHG
+
+/*
+ * Mininum number of partial slabs. These will be left on the partial
+ * lists even if they are empty. kmem_cache_shrink may reclaim them.
+ */
+#define MIN_PARTIAL 5
+
+/*
+ * Maximum number of desirable partial slabs.
+ * The existence of more partial slabs makes kmem_cache_shrink
+ * sort the partial list by the number of objects in the.
+ */
+#define MAX_PARTIAL 10
+
+#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
+                SLAB_POISON | SLAB_STORE_USER)
+
+/*
+ * Debugging flags that require metadata to be stored in the slab.  These get
+ * disabled when slub_debug=O is used and a cache's min order increases with
+ * metadata.
+ */
+#define DEBUG_METADATA_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
+
+/*
+ * 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_FAILSLAB)
+
+#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
+        SLAB_CACHE_DMA | SLAB_NOTRACK)
+
+#define OO_SHIFT    16
+#define OO_MASK        ((1 << OO_SHIFT) - 1)
+#define MAX_OBJS_PER_PAGE    32767 /* since page.objects is u15 */
+
+/* Internal SLUB flags */
+#define __OBJECT_POISON        0x80000000UL /* Poison object */
+#define __CMPXCHG_DOUBLE    0x40000000UL /* Use cmpxchg_double */
+
+#ifdef CONFIG_SMP
+static struct notifier_block slab_notifier;
+#endif
+
+/*
+ * Tracking user of a slab.
+ */
+#define TRACK_ADDRS_COUNT 16
+struct track {
+    unsigned long addr;    /* Called from address */
+#ifdef CONFIG_STACKTRACE
+    unsigned long addrs[TRACK_ADDRS_COUNT];    /* Called from address */
+#endif
+    int cpu;        /* Was running on cpu */
+    int pid;        /* Pid context */
+    unsigned long when;    /* When did the operation occur */
+};
+
+enum track_item { TRACK_ALLOC, TRACK_FREE };
+
+#ifdef CONFIG_SYSFS
+static int sysfs_slab_add(struct kmem_cache *);
+static int sysfs_slab_alias(struct kmem_cache *, const char *);
+static void sysfs_slab_remove(struct kmem_cache *);
+static void memcg_propagate_slab_attrs(struct kmem_cache *s);
+#else
+static inline int sysfs_slab_add(struct kmem_cache *s) { return 0; }
+static inline int sysfs_slab_alias(struct kmem_cache *s, const char *p)
+                            { return 0; }
+static inline void sysfs_slab_remove(struct kmem_cache *s) { }
+
+static inline void memcg_propagate_slab_attrs(struct kmem_cache *s) { }
+#endif
+
+static inline void stat(const struct kmem_cache *s, enum stat_item si)
+{
+#ifdef CONFIG_SLEB_STATS
+    __this_cpu_inc(s->cpu_slab->stat[si]);
+#endif
+}
+
+/********************************************************************
+ *             Core slab cache functions
+ *******************************************************************/
+
+static inline struct kmem_cache_node *get_node(struct kmem_cache *s, int node)
+{
+    return s->node[node];
+}
+
+/* Verify that a pointer has an address that is valid within a slab page */
+static inline int check_valid_pointer(struct kmem_cache *s,
+                struct page *page, const void *object)
+{
+    void *base;
+
+    if (!object)
+        return 1;
+
+    base = page_address(page);
+    if (object < base || object >= base + page->objects * s->size ||
+        (object - base) % s->size) {
+        return 0;
+    }
+
+    return 1;
+}
+
+static inline void *get_freepointer(struct kmem_cache *s, void *object)
+{
+    return *(void **)(object + s->offset);
+}
+
+static void prefetch_freepointer(const struct kmem_cache *s, void *object)
+{
+    prefetch(object + s->offset);
+}
+
+static inline void *get_freepointer_safe(struct kmem_cache *s, void *object)
+{
+    void *p;
+
+#ifdef CONFIG_DEBUG_PAGEALLOC
+    probe_kernel_read(&p, (void **)(object + s->offset), sizeof(p));
+#else
+    p = get_freepointer(s, object);
+#endif
+    return p;
+}
+
+static inline void set_freepointer(struct kmem_cache *s, void *object, void *fp)
+{
+    *(void **)(object + s->offset) = fp;
+}
+
+/* Loop over all objects in a slab */
+#define for_each_object(__p, __s, __addr, __objects) \
+    for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
+            __p += (__s)->size)
+
+/* Determine object index from a given position */
+static inline int slab_index(void *p, struct kmem_cache *s, void *addr)
+{
+    return (p - addr) / s->size;
+}
+
+static inline size_t slab_ksize(const struct kmem_cache *s)
+{
+#ifdef CONFIG_SLEB_DEBUG
+    /*
+     * Debugging requires use of the padding between object
+     * and whatever may come after it.
+     */
+    if (s->flags & (SLAB_RED_ZONE | SLAB_POISON))
+        return s->object_size;
+
+#endif
+    /*
+     * If we have the need to store the freelist pointer
+     * back there or track user information then we can
+     * only use the space before that information.
+     */
+    if (s->flags & (SLAB_DESTROY_BY_RCU | SLAB_STORE_USER))
+        return s->inuse;
+    /*
+     * Else we can use all the padding etc for the allocation
+     */
+    return s->size;
+}
+
+static inline int order_objects(int order, unsigned long size, int reserved)
+{
+    return ((PAGE_SIZE << order) - reserved) / size;
+}
+
+static inline struct kmem_cache_order_objects oo_make(int order,
+        unsigned long size, int reserved)
+{
+    struct kmem_cache_order_objects x = {
+        (order << OO_SHIFT) + order_objects(order, size, reserved)
+    };
+
+    return x;
+}
+
+static inline int oo_order(struct kmem_cache_order_objects x)
+{
+    return x.x >> OO_SHIFT;
+}
+
+static inline int oo_objects(struct kmem_cache_order_objects x)
+{
+    return x.x & OO_MASK;
+}
+
+/*
+ * Per slab locking using the pagelock
+ */
+static __always_inline void slab_lock(struct page *page)
+{
+    bit_spin_lock(PG_locked, &page->flags);
+}
+
+static __always_inline void slab_unlock(struct page *page)
+{
+    __bit_spin_unlock(PG_locked, &page->flags);
+}
+
+/* Interrupts must be disabled (for the fallback code to work right) */
+static inline bool __cmpxchg_double_slab(struct kmem_cache *s, struct page *page,
+        void *freelist_old, unsigned long counters_old,
+        void *freelist_new, unsigned long counters_new,
+        const char *n)
+{
+    VM_BUG_ON(!irqs_disabled());
+#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
+    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
+    if (s->flags & __CMPXCHG_DOUBLE) {
+        if (cmpxchg_double(&page->freelist, &page->counters,
+            freelist_old, counters_old,
+            freelist_new, counters_new))
+        return 1;
+    } else
+#endif
+    {
+        slab_lock(page);
+        if (page->freelist == freelist_old && page->counters == counters_old) {
+            page->freelist = freelist_new;
+            page->counters = counters_new;
+            slab_unlock(page);
+            return 1;
+        }
+        slab_unlock(page);
+    }
+
+    cpu_relax();
+    stat(s, CMPXCHG_DOUBLE_FAIL);
+
+#ifdef SLUB_DEBUG_CMPXCHG
+    printk(KERN_INFO "%s %s: cmpxchg double redo ", n, s->name);
+#endif
+
+    return 0;
+}
+
+static inline bool cmpxchg_double_slab(struct kmem_cache *s, struct page *page,
+        void *freelist_old, unsigned long counters_old,
+        void *freelist_new, unsigned long counters_new,
+        const char *n)
+{
+#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
+    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
+    if (s->flags & __CMPXCHG_DOUBLE) {
+        if (cmpxchg_double(&page->freelist, &page->counters,
+            freelist_old, counters_old,
+            freelist_new, counters_new))
+        return 1;
+    } else
+#endif
+    {
+        unsigned long flags;
+
+        local_irq_save(flags);
+        slab_lock(page);
+        if (page->freelist == freelist_old && page->counters == counters_old) {
+            page->freelist = freelist_new;
+            page->counters = counters_new;
+            slab_unlock(page);
+            local_irq_restore(flags);
+            return 1;
+        }
+        slab_unlock(page);
+        local_irq_restore(flags);
+    }
+
+    cpu_relax();
+    stat(s, CMPXCHG_DOUBLE_FAIL);
+
+#ifdef SLUB_DEBUG_CMPXCHG
+    printk(KERN_INFO "%s %s: cmpxchg double redo ", n, s->name);
+#endif
+
+    return 0;
+}
+
+#ifdef CONFIG_SLEB_DEBUG
+/*
+ * Determine a map of object in use on a page.
+ *
+ * Node listlock must be held to guarantee that the page does
+ * not vanish from under us.
+ */
+static void get_map(struct kmem_cache *s, struct page *page, unsigned long *map)
+{
+    void *p;
+    void *addr = page_address(page);
+
+    for (p = page->freelist; p; p = get_freepointer(s, p))
+        set_bit(slab_index(p, s, addr), map);
+}
+
+/*
+ * Debug settings:
+ */
+#ifdef CONFIG_SLEB_DEBUG_ON
+static int slub_debug = DEBUG_DEFAULT_FLAGS;
+#else
+static int slub_debug;
+#endif
+
+static char *slub_debug_slabs;
+static int disable_higher_order_debug;
+
+/*
+ * Object debugging
+ */
+static void print_section(char *text, u8 *addr, unsigned int length)
+{
+    print_hex_dump(KERN_ERR, text, DUMP_PREFIX_ADDRESS, 16, 1, addr,
+            length, 1);
+}
+
+static struct track *get_track(struct kmem_cache *s, void *object,
+    enum track_item alloc)
+{
+    struct track *p;
+
+    if (s->offset)
+        p = object + s->offset + sizeof(void *);
+    else
+        p = object + s->inuse;
+
+    return p + alloc;
+}
+
+static void set_track(struct kmem_cache *s, void *object,
+            enum track_item alloc, unsigned long addr)
+{
+    struct track *p = get_track(s, object, alloc);
+
+    if (addr) {
+#ifdef CONFIG_STACKTRACE
+        struct stack_trace trace;
+        int i;
+
+        trace.nr_entries = 0;
+        trace.max_entries = TRACK_ADDRS_COUNT;
+        trace.entries = p->addrs;
+        trace.skip = 3;
+        save_stack_trace(&trace);
+
+        /* See rant in lockdep.c */
+        if (trace.nr_entries != 0 &&
+            trace.entries[trace.nr_entries - 1] == ULONG_MAX)
+            trace.nr_entries--;
+
+        for (i = trace.nr_entries; i < TRACK_ADDRS_COUNT; i++)
+            p->addrs[i] = 0;
+#endif
+        p->addr = addr;
+        p->cpu = smp_processor_id();
+        p->pid = current->pid;
+        p->when = jiffies;
+    } else
+        memset(p, 0, sizeof(struct track));
+}
+
+static void init_tracking(struct kmem_cache *s, void *object)
+{
+    if (!(s->flags & SLAB_STORE_USER))
+        return;
+
+    set_track(s, object, TRACK_FREE, 0UL);
+    set_track(s, object, TRACK_ALLOC, 0UL);
+}
+
+static void print_track(const char *s, struct track *t)
+{
+    if (!t->addr)
+        return;
+
+    printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
+        s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
+#ifdef CONFIG_STACKTRACE
+    {
+        int i;
+        for (i = 0; i < TRACK_ADDRS_COUNT; i++)
+            if (t->addrs[i])
+                printk(KERN_ERR "\t%pS\n", (void *)t->addrs[i]);
+            else
+                break;
+    }
+#endif
+}
+
+static void print_tracking(struct kmem_cache *s, void *object)
+{
+    if (!(s->flags & SLAB_STORE_USER))
+        return;
+
+    print_track("Allocated", get_track(s, object, TRACK_ALLOC));
+    print_track("Freed", get_track(s, object, TRACK_FREE));
+}
+
+static void print_page_info(struct page *page)
+{
+    printk(KERN_ERR "INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=0x%04lx\n",
+        page, page->objects, page->inuse, page->freelist, page->flags);
+
+}
+
+static void slab_bug(struct kmem_cache *s, char *fmt, ...)
+{
+    va_list args;
+    char buf[100];
+
+    va_start(args, fmt);
+    vsnprintf(buf, sizeof(buf), fmt, args);
+    va_end(args);
+    printk(KERN_ERR "========================================"
+            "=====================================\n");
+    printk(KERN_ERR "BUG %s (%s): %s\n", s->name, print_tainted(), buf);
+    printk(KERN_ERR "----------------------------------------"
+            "-------------------------------------\n\n");
+
+    add_taint(TAINT_BAD_PAGE, LOCKDEP_NOW_UNRELIABLE);
+}
+
+static void slab_fix(struct kmem_cache *s, char *fmt, ...)
+{
+    va_list args;
+    char buf[100];
+
+    va_start(args, fmt);
+    vsnprintf(buf, sizeof(buf), fmt, args);
+    va_end(args);
+    printk(KERN_ERR "FIX %s: %s\n", s->name, buf);
+}
+
+static void print_trailer(struct kmem_cache *s, struct page *page, u8 *p)
+{
+    unsigned int off;    /* Offset of last byte */
+    u8 *addr = page_address(page);
+
+    print_tracking(s, p);
+
+    print_page_info(page);
+
+    printk(KERN_ERR "INFO: Object 0x%p @offset=%tu fp=0x%p\n\n",
+            p, p - addr, get_freepointer(s, p));
+
+    if (p > addr + 16)
+        print_section("Bytes b4 ", p - 16, 16);
+
+    print_section("Object ", p, min_t(unsigned long, s->object_size,
+                PAGE_SIZE));
+    if (s->flags & SLAB_RED_ZONE)
+        print_section("Redzone ", p + s->object_size,
+            s->inuse - s->object_size);
+
+    if (s->offset)
+        off = s->offset + sizeof(void *);
+    else
+        off = s->inuse;
+
+    if (s->flags & SLAB_STORE_USER)
+        off += 2 * sizeof(struct track);
+
+    if (off != s->size)
+        /* Beginning of the filler is the free pointer */
+        print_section("Padding ", p + off, s->size - off);
+
+    dump_stack();
+}
+
+static void object_err(struct kmem_cache *s, struct page *page,
+            u8 *object, char *reason)
+{
+    slab_bug(s, "%s", reason);
+    print_trailer(s, page, object);
+}
+
+static void slab_err(struct kmem_cache *s, struct page *page, const char *fmt, ...)
+{
+    va_list args;
+    char buf[100];
+
+    va_start(args, fmt);
+    vsnprintf(buf, sizeof(buf), fmt, args);
+    va_end(args);
+    slab_bug(s, "%s", buf);
+    print_page_info(page);
+    dump_stack();
+}
+
+static void init_object(struct kmem_cache *s, void *object, u8 val)
+{
+    u8 *p = object;
+
+    if (s->flags & __OBJECT_POISON) {
+        memset(p, POISON_FREE, s->object_size - 1);
+        p[s->object_size - 1] = POISON_END;
+    }
+
+    if (s->flags & SLAB_RED_ZONE)
+        memset(p + s->object_size, val, s->inuse - s->object_size);
+}
+
+static void restore_bytes(struct kmem_cache *s, char *message, u8 data,
+                        void *from, void *to)
+{
+    slab_fix(s, "Restoring 0x%p-0x%p=0x%x\n", from, to - 1, data);
+    memset(from, data, to - from);
+}
+
+static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
+            u8 *object, char *what,
+            u8 *start, unsigned int value, unsigned int bytes)
+{
+    u8 *fault;
+    u8 *end;
+
+    fault = memchr_inv(start, value, bytes);
+    if (!fault)
+        return 1;
+
+    end = start + bytes;
+    while (end > fault && end[-1] == value)
+        end--;
+
+    slab_bug(s, "%s overwritten", what);
+    printk(KERN_ERR "INFO: 0x%p-0x%p. First byte 0x%x instead of 0x%x\n",
+                    fault, end - 1, fault[0], value);
+    print_trailer(s, page, object);
+
+    restore_bytes(s, what, value, fault, end);
+    return 0;
+}
+
+/*
+ * Object layout:
+ *
+ * object address
+ *     Bytes of the object to be managed.
+ *     If the freepointer may overlay the object then the free
+ *     pointer is the first word of the object.
+ *
+ *     Poisoning uses 0x6b (POISON_FREE) and the last byte is
+ *     0xa5 (POISON_END)
+ *
+ * object + s->object_size
+ *     Padding to reach word boundary. This is also used for Redzoning.
+ *     Padding is extended by another word if Redzoning is enabled and
+ *     object_size == inuse.
+ *
+ *     We fill with 0xbb (RED_INACTIVE) for inactive objects and with
+ *     0xcc (RED_ACTIVE) for objects in use.
+ *
+ * object + s->inuse
+ *     Meta data starts here.
+ *
+ *     A. Free pointer (if we cannot overwrite object on free)
+ *     B. Tracking data for SLAB_STORE_USER
+ *     C. Padding to reach required alignment boundary or at mininum
+ *         one word if debugging is on to be able to detect writes
+ *         before the word boundary.
+ *
+ *    Padding is done using 0x5a (POISON_INUSE)
+ *
+ * object + s->size
+ *     Nothing is used beyond s->size.
+ *
+ * If slabcaches are merged then the object_size and inuse boundaries are mostly
+ * ignored. And therefore no slab options that rely on these boundaries
+ * may be used with merged slabcaches.
+ */
+
+static int check_pad_bytes(struct kmem_cache *s, struct page *page, u8 *p)
+{
+    unsigned long off = s->inuse;    /* The end of info */
+
+    if (s->offset)
+        /* Freepointer is placed after the object. */
+        off += sizeof(void *);
+
+    if (s->flags & SLAB_STORE_USER)
+        /* We also have user information there */
+        off += 2 * sizeof(struct track);
+
+    if (s->size == off)
+        return 1;
+
+    return check_bytes_and_report(s, page, p, "Object padding",
+                p + off, POISON_INUSE, s->size - off);
+}
+
+/* Check the pad bytes at the end of a slab page */
+static int slab_pad_check(struct kmem_cache *s, struct page *page)
+{
+    u8 *start;
+    u8 *fault;
+    u8 *end;
+    int length;
+    int remainder;
+
+    if (!(s->flags & SLAB_POISON))
+        return 1;
+
+    start = page_address(page);
+    length = (PAGE_SIZE << compound_order(page)) - s->reserved;
+    end = start + length;
+    remainder = length % s->size;
+    if (!remainder)
+        return 1;
+
+    fault = memchr_inv(end - remainder, POISON_INUSE, remainder);
+    if (!fault)
+        return 1;
+    while (end > fault && end[-1] == POISON_INUSE)
+        end--;
+
+    slab_err(s, page, "Padding overwritten. 0x%p-0x%p", fault, end - 1);
+    print_section("Padding ", end - remainder, remainder);
+
+    restore_bytes(s, "slab padding", POISON_INUSE, end - remainder, end);
+    return 0;
+}
+
+static int check_object(struct kmem_cache *s, struct page *page,
+                    void *object, u8 val)
+{
+    u8 *p = object;
+    u8 *endobject = object + s->object_size;
+
+    if (s->flags & SLAB_RED_ZONE) {
+        if (!check_bytes_and_report(s, page, object, "Redzone",
+            endobject, val, s->inuse - s->object_size))
+            return 0;
+    } else {
+        if ((s->flags & SLAB_POISON) && s->object_size < s->inuse) {
+            check_bytes_and_report(s, page, p, "Alignment padding",
+                endobject, POISON_INUSE, s->inuse - s->object_size);
+        }
+    }
+
+    if (s->flags & SLAB_POISON) {
+        if (val != SLUB_RED_ACTIVE && (s->flags & __OBJECT_POISON) &&
+            (!check_bytes_and_report(s, page, p, "Poison", p,
+                    POISON_FREE, s->object_size - 1) ||
+             !check_bytes_and_report(s, page, p, "Poison",
+                p + s->object_size - 1, POISON_END, 1)))
+            return 0;
+        /*
+         * check_pad_bytes cleans up on its own.
+         */
+        check_pad_bytes(s, page, p);
+    }
+
+    if (!s->offset && val == SLUB_RED_ACTIVE)
+        /*
+         * Object and freepointer overlap. Cannot check
+         * freepointer while object is allocated.
+         */
+        return 1;
+
+    /* Check free pointer validity */
+    if (!check_valid_pointer(s, page, get_freepointer(s, p))) {
+        object_err(s, page, p, "Freepointer corrupt");
+        /*
+         * No choice but to zap it and thus lose the remainder
+         * of the free objects in this slab. May cause
+         * another error because the object count is now wrong.
+         */
+        set_freepointer(s, p, NULL);
+        return 0;
+    }
+    return 1;
+}
+
+static int check_slab(struct kmem_cache *s, struct page *page)
+{
+    int maxobj;
+
+    VM_BUG_ON(!irqs_disabled());
+
+    if (!PageSlab(page)) {
+        slab_err(s, page, "Not a valid slab page");
+        return 0;
+    }
+
+    maxobj = order_objects(compound_order(page), s->size, s->reserved);
+    if (page->objects > maxobj) {
+        slab_err(s, page, "objects %u > max %u",
+            s->name, page->objects, maxobj);
+        return 0;
+    }
+    if (page->inuse > page->objects) {
+        slab_err(s, page, "inuse %u > max %u",
+            s->name, page->inuse, page->objects);
+        return 0;
+    }
+    /* Slab_pad_check fixes things up after itself */
+    slab_pad_check(s, page);
+    return 1;
+}
+
+/*
+ * Determine if a certain object on a page is on the freelist. Must hold the
+ * slab lock to guarantee that the chains are in a consistent state.
+ */
+static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
+{
+    int nr = 0;
+    void *fp;
+    void *object = NULL;
+    unsigned long max_objects;
+
+    fp = page->freelist;
+    while (fp && nr <= page->objects) {
+        if (fp == search)
+            return 1;
+        if (!check_valid_pointer(s, page, fp)) {
+            if (object) {
+                object_err(s, page, object,
+                    "Freechain corrupt");
+                set_freepointer(s, object, NULL);
+                break;
+            } else {
+                slab_err(s, page, "Freepointer corrupt");
+                page->freelist = NULL;
+                page->inuse = page->objects;
+                slab_fix(s, "Freelist cleared");
+                return 0;
+            }
+            break;
+        }
+        object = fp;
+        fp = get_freepointer(s, object);
+        nr++;
+    }
+
+    max_objects = order_objects(compound_order(page), s->size, s->reserved);
+    if (max_objects > MAX_OBJS_PER_PAGE)
+        max_objects = MAX_OBJS_PER_PAGE;
+
+    if (page->objects != max_objects) {
+        slab_err(s, page, "Wrong number of objects. Found %d but "
+            "should be %d", page->objects, max_objects);
+        page->objects = max_objects;
+        slab_fix(s, "Number of objects adjusted.");
+    }
+    if (page->inuse != page->objects - nr) {
+        slab_err(s, page, "Wrong object count. Counter is %d but "
+            "counted were %d", page->inuse, page->objects - nr);
+        page->inuse = page->objects - nr;
+        slab_fix(s, "Object count adjusted.");
+    }
+    return search == NULL;
+}
+
+static void trace(struct kmem_cache *s, struct page *page, void *object,
+                                int alloc)
+{
+    if (s->flags & SLAB_TRACE) {
+        printk(KERN_INFO "TRACE %s %s 0x%p inuse=%d fp=0x%p\n",
+            s->name,
+            alloc ? "alloc" : "free",
+            object, page->inuse,
+            page->freelist);
+
+        if (!alloc)
+            print_section("Object ", (void *)object, s->object_size);
+
+        dump_stack();
+    }
+}
+
+/*
+ * Hooks for other subsystems that check memory allocations. In a typical
+ * production configuration these hooks all should produce no code at all.
+ */
+static inline int slab_pre_alloc_hook(struct kmem_cache *s, gfp_t flags)
+{
+    flags &= gfp_allowed_mask;
+    lockdep_trace_alloc(flags);
+    might_sleep_if(flags & __GFP_WAIT);
+
+    return should_failslab(s->object_size, flags, s->flags);
+}
+
+static inline void slab_post_alloc_hook(struct kmem_cache *s, gfp_t flags, void *object)
+{
+    flags &= gfp_allowed_mask;
+    kmemcheck_slab_alloc(s, flags, object, slab_ksize(s));
+    kmemleak_alloc_recursive(object, s->object_size, 1, s->flags, flags);
+}
+
+static inline void slab_free_hook(struct kmem_cache *s, void *x)
+{
+    kmemleak_free_recursive(x, s->flags);
+
+    /*
+     * Trouble is that we may no longer disable interupts in the fast path
+     * So in order to make the debug calls that expect irqs to be
+     * disabled we need to disable interrupts temporarily.
+     */
+#if defined(CONFIG_KMEMCHECK) || defined(CONFIG_LOCKDEP)
+    {
+        unsigned long flags;
+
+        local_irq_save(flags);
+        kmemcheck_slab_free(s, x, s->object_size);
+        debug_check_no_locks_freed(x, s->object_size);
+        local_irq_restore(flags);
+    }
+#endif
+    if (!(s->flags & SLAB_DEBUG_OBJECTS))
+        debug_check_no_obj_freed(x, s->object_size);
+}
+
+/*
+ * Tracking of fully allocated slabs for debugging purposes.
+ *
+ * list_lock must be held.
+ */
+static void add_full(struct kmem_cache *s,
+    struct kmem_cache_node *n, struct page *page)
+{
+    if (!(s->flags & SLAB_STORE_USER))
+        return;
+
+    list_add(&page->lru, &n->full);
+}
+
+/*
+ * list_lock must be held.
+ */
+static void remove_full(struct kmem_cache *s, struct page *page)
+{
+    if (!(s->flags & SLAB_STORE_USER))
+        return;
+
+    list_del(&page->lru);
+}
+
+/* Tracking of the number of slabs for debugging purposes */
+static inline unsigned long slabs_node(struct kmem_cache *s, int node)
+{
+    struct kmem_cache_node *n = get_node(s, node);
+
+    return atomic_long_read(&n->nr_slabs);
+}
+
+static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
+{
+    return atomic_long_read(&n->nr_slabs);
+}
+
+static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
+{
+    struct kmem_cache_node *n = get_node(s, node);
+
+    /*
+     * May be called early in order to allocate a slab for the
+     * kmem_cache_node structure. Solve the chicken-egg
+     * dilemma by deferring the increment of the count during
+     * bootstrap (see early_kmem_cache_node_alloc).
+     */
+    if (likely(n)) {
+        atomic_long_inc(&n->nr_slabs);
+        atomic_long_add(objects, &n->total_objects);
+    }
+}
+static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
+{
+    struct kmem_cache_node *n = get_node(s, node);
+
+    atomic_long_dec(&n->nr_slabs);
+    atomic_long_sub(objects, &n->total_objects);
+}
+
+/* Object debug checks for alloc/free paths */
+static void setup_object_debug(struct kmem_cache *s, struct page *page,
+                                void *object)
+{
+    if (!(s->flags & (SLAB_STORE_USER|SLAB_RED_ZONE|__OBJECT_POISON)))
+        return;
+
+    init_object(s, object, SLUB_RED_INACTIVE);
+    init_tracking(s, object);
+}
+
+static noinline int alloc_debug_processing(struct kmem_cache *s, struct page *page,
+                    void *object, unsigned long addr)
+{
+    if (!check_slab(s, page))
+        goto bad;
+
+    if (!check_valid_pointer(s, page, object)) {
+        object_err(s, page, object, "Freelist Pointer check fails");
+        goto bad;
+    }
+
+    if (!check_object(s, page, object, SLUB_RED_INACTIVE))
+        goto bad;
+
+    /* Success perform special debug activities for allocs */
+    if (s->flags & SLAB_STORE_USER)
+        set_track(s, object, TRACK_ALLOC, addr);
+    trace(s, page, object, 1);
+    init_object(s, object, SLUB_RED_ACTIVE);
+    return 1;
+
+bad:
+    if (PageSlab(page)) {
+        /*
+         * If this is a slab page then lets do the best we can
+         * to avoid issues in the future. Marking all objects
+         * as used avoids touching the remaining objects.
+         */
+        slab_fix(s, "Marking all objects used");
+        page->inuse = page->objects;
+        page->freelist = NULL;
+    }
+    return 0;
+}
+
+static noinline struct kmem_cache_node *free_debug_processing(
+    struct kmem_cache *s, struct page *page, void *object,
+    unsigned long addr, unsigned long *flags)
+{
+    struct kmem_cache_node *n = get_node(s, page_to_nid(page));
+
+    spin_lock_irqsave(&n->list_lock, *flags);
+    slab_lock(page);
+
+    if (!check_slab(s, page))
+        goto fail;
+
+    if (!check_valid_pointer(s, page, object)) {
+        slab_err(s, page, "Invalid object pointer 0x%p", object);
+        goto fail;
+    }
+
+    if (on_freelist(s, page, object)) {
+        object_err(s, page, object, "Object already free");
+        goto fail;
+    }
+
+    if (!check_object(s, page, object, SLUB_RED_ACTIVE))
+        goto out;
+
+    if (unlikely(s != page->slab_cache)) {
+        if (!PageSlab(page)) {
+            slab_err(s, page, "Attempt to free object(0x%p) "
+                "outside of slab", object);
+        } else if (!page->slab_cache) {
+            printk(KERN_ERR
+                "SLUB <none>: no slab for object 0x%p.\n",
+                        object);
+            dump_stack();
+        } else
+            object_err(s, page, object,
+                    "page slab pointer corrupt.");
+        goto fail;
+    }
+
+    if (s->flags & SLAB_STORE_USER)
+        set_track(s, object, TRACK_FREE, addr);
+    trace(s, page, object, 0);
+    init_object(s, object, SLUB_RED_INACTIVE);
+out:
+    slab_unlock(page);
+    /*
+     * Keep node_lock to preserve integrity
+     * until the object is actually freed
+     */
+    return n;
+
+fail:
+    slab_unlock(page);
+    spin_unlock_irqrestore(&n->list_lock, *flags);
+    slab_fix(s, "Object at 0x%p not freed", object);
+    return NULL;
+}
+
+static int __init setup_slub_debug(char *str)
+{
+    slub_debug = DEBUG_DEFAULT_FLAGS;
+    if (*str++ != '=' || !*str)
+        /*
+         * No options specified. Switch on full debugging.
+         */
+        goto out;
+
+    if (*str == ',')
+        /*
+         * No options but restriction on slabs. This means full
+         * debugging for slabs matching a pattern.
+         */
+        goto check_slabs;
+
+    if (tolower(*str) == 'o') {
+        /*
+         * Avoid enabling debugging on caches if its minimum order
+         * would increase as a result.
+         */
+        disable_higher_order_debug = 1;
+        goto out;
+    }
+
+    slub_debug = 0;
+    if (*str == '-')
+        /*
+         * Switch off all debugging measures.
+         */
+        goto out;
+
+    /*
+     * Determine which debug features should be switched on
+     */
+    for (; *str && *str != ','; str++) {
+        switch (tolower(*str)) {
+        case 'f':
+            slub_debug |= SLAB_DEBUG_FREE;
+            break;
+        case 'z':
+            slub_debug |= SLAB_RED_ZONE;
+            break;
+        case 'p':
+            slub_debug |= SLAB_POISON;
+            break;
+        case 'u':
+            slub_debug |= SLAB_STORE_USER;
+            break;
+        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);
+        }
+    }
+
+check_slabs:
+    if (*str == ',')
+        slub_debug_slabs = str + 1;
+out:
+    return 1;
+}
+
+__setup("slub_debug", setup_slub_debug);
+
+static unsigned long kmem_cache_flags(unsigned long object_size,
+    unsigned long flags, const char *name,
+    void (*ctor)(void *))
+{
+    /*
+     * Enable debugging if selected on the kernel commandline.
+     */
+    if (slub_debug && (!slub_debug_slabs ||
+        !strncmp(slub_debug_slabs, name, strlen(slub_debug_slabs))))
+        flags |= slub_debug;
+
+    return flags;
+}
+#else    /*!CONFIG_SLEB_DEBUG*/
+static inline void setup_object_debug(struct kmem_cache *s,
+            struct page *page, void *object) {}
+
+static inline int alloc_debug_processing(struct kmem_cache *s,
+    struct page *page, void *object, unsigned long addr) { return 0; }
+
+static inline struct kmem_cache_node *free_debug_processing(
+    struct kmem_cache *s, struct page *page, void *object,
+    unsigned long addr, unsigned long *flags) { return NULL; }
+
+static inline int slab_pad_check(struct kmem_cache *s, struct page *page)
+            { return 1; }
+static inline int check_object(struct kmem_cache *s, struct page *page,
+            void *object, u8 val) { return 1; }
+static inline void add_full(struct kmem_cache *s, struct kmem_cache_node *n,
+                    struct page *page) {}
+static inline void remove_full(struct kmem_cache *s, struct page *page) {}
+static inline unsigned long kmem_cache_flags(unsigned long object_size,
+    unsigned long flags, const char *name,
+    void (*ctor)(void *))
+{
+    return flags;
+}
+#define slub_debug 0
+
+#define disable_higher_order_debug 0
+
+static inline unsigned long slabs_node(struct kmem_cache *s, int node)
+                            { return 0; }
+static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
+                            { return 0; }
+static inline void inc_slabs_node(struct kmem_cache *s, int node,
+                            int objects) {}
+static inline void dec_slabs_node(struct kmem_cache *s, int node,
+                            int objects) {}
+
+static inline int slab_pre_alloc_hook(struct kmem_cache *s, gfp_t flags)
+                            { return 0; }
+
+static inline void slab_post_alloc_hook(struct kmem_cache *s, gfp_t flags,
+        void *object) {}
+
+static inline void slab_free_hook(struct kmem_cache *s, void *x) {}
+
+#endif /* CONFIG_SLEB_DEBUG */
+
+/*
+ * Slab allocation and freeing
+ */
+static inline struct page *alloc_slab_page(gfp_t flags, int node,
+                    struct kmem_cache_order_objects oo)
+{
+    int order = oo_order(oo);
+
+    flags |= __GFP_NOTRACK;
+
+    if (node == NUMA_NO_NODE)
+        return alloc_pages(flags, order);
+    else
+        return alloc_pages_exact_node(node, flags, order);
+}
+
+static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
+{
+    struct page *page;
+    struct kmem_cache_order_objects oo = s->oo;
+    gfp_t alloc_gfp;
+
+    flags &= gfp_allowed_mask;
+
+    if (flags & __GFP_WAIT)
+        local_irq_enable();
+
+    flags |= s->allocflags;
+
+    /*
+     * Let the initial higher-order allocation fail under memory pressure
+     * so we fall-back to the minimum order allocation.
+     */
+    alloc_gfp = (flags | __GFP_NOWARN | __GFP_NORETRY) & ~__GFP_NOFAIL;
+
+    page = alloc_slab_page(alloc_gfp, node, oo);
+    if (unlikely(!page)) {
+        oo = s->min;
+        /*
+         * Allocation may have failed due to fragmentation.
+         * Try a lower order alloc if possible
+         */
+        page = alloc_slab_page(flags, node, oo);
+
+        if (page)
+            stat(s, ORDER_FALLBACK);
+    }
+
+    if (kmemcheck_enabled && page
+        && !(s->flags & (SLAB_NOTRACK | DEBUG_DEFAULT_FLAGS))) {
+        int pages = 1 << oo_order(oo);
+
+        kmemcheck_alloc_shadow(page, oo_order(oo), flags, node);
+
+        /*
+         * Objects from caches that have a constructor don't get
+         * cleared when they're allocated, so we need to do it here.
+         */
+        if (s->ctor)
+            kmemcheck_mark_uninitialized_pages(page, pages);
+        else
+            kmemcheck_mark_unallocated_pages(page, pages);
+    }
+
+    if (flags & __GFP_WAIT)
+        local_irq_disable();
+    if (!page)
+        return NULL;
+
+    page->objects = oo_objects(oo);
+    mod_zone_page_state(page_zone(page),
+        (s->flags & SLAB_RECLAIM_ACCOUNT) ?
+        NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
+        1 << oo_order(oo));
+
+    return page;
+}
+
+static void setup_object(struct kmem_cache *s, struct page *page,
+                void *object)
+{
+    setup_object_debug(s, page, object);
+    if (unlikely(s->ctor))
+        s->ctor(object);
+}
+
+static struct page *new_slab(struct kmem_cache *s, gfp_t flags, int node)
+{
+    struct page *page;
+    void *start;
+    void *last;
+    void *p;
+    int order;
+
+    BUG_ON(flags & GFP_SLAB_BUG_MASK);
+
+    page = allocate_slab(s,
+        flags & (GFP_RECLAIM_MASK | GFP_CONSTRAINT_MASK), node);
+    if (!page)
+        goto out;
+
+    order = compound_order(page);
+    inc_slabs_node(s, page_to_nid(page), page->objects);
+    memcg_bind_pages(s, order);
+    page->slab_cache = s;
+    __SetPageSlab(page);
+    if (page->pfmemalloc)
+        SetPageSlabPfmemalloc(page);
+
+    start = page_address(page);
+
+    if (unlikely(s->flags & SLAB_POISON))
+        memset(start, POISON_INUSE, PAGE_SIZE << order);
+
+    last = start;
+    for_each_object(p, s, start, page->objects) {
+        setup_object(s, page, last);
+        set_freepointer(s, last, p);
+        last = p;
+    }
+    setup_object(s, page, last);
+    set_freepointer(s, last, NULL);
+
+    page->freelist = start;
+    page->inuse = page->objects;
+    page->frozen = 1;
+out:
+    return page;
+}
+
+static void __free_slab(struct kmem_cache *s, struct page *page)
+{
+    int order = compound_order(page);
+    int pages = 1 << order;
+
+    if (kmem_cache_debug(s)) {
+        void *p;
+
+        slab_pad_check(s, page);
+        for_each_object(p, s, page_address(page),
+                        page->objects)
+            check_object(s, page, p, SLUB_RED_INACTIVE);
+    }
+
+    kmemcheck_free_shadow(page, compound_order(page));
+
+    mod_zone_page_state(page_zone(page),
+        (s->flags & SLAB_RECLAIM_ACCOUNT) ?
+        NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
+        -pages);
+
+    __ClearPageSlabPfmemalloc(page);
+    __ClearPageSlab(page);
+
+    memcg_release_pages(s, order);
+    page_mapcount_reset(page);
+    if (current->reclaim_state)
+        current->reclaim_state->reclaimed_slab += pages;
+    __free_memcg_kmem_pages(page, order);
+}
+
+#define need_reserve_slab_rcu                        \
+    (sizeof(((struct page *)NULL)->lru) < sizeof(struct rcu_head))
+
+static void rcu_free_slab(struct rcu_head *h)
+{
+    struct page *page;
+
+    if (need_reserve_slab_rcu)
+        page = virt_to_head_page(h);
+    else
+        page = container_of((struct list_head *)h, struct page, lru);
+
+    __free_slab(page->slab_cache, page);
+}
+
+static void free_slab(struct kmem_cache *s, struct page *page)
+{
+    if (unlikely(s->flags & SLAB_DESTROY_BY_RCU)) {
+        struct rcu_head *head;
+
+        if (need_reserve_slab_rcu) {
+            int order = compound_order(page);
+            int offset = (PAGE_SIZE << order) - s->reserved;
+
+            VM_BUG_ON(s->reserved != sizeof(*head));
+            head = page_address(page) + offset;
+        } else {
+            /*
+             * RCU free overloads the RCU head over the LRU
+             */
+            head = (void *)&page->lru;
+        }
+
+        call_rcu(head, rcu_free_slab);
+    } else
+        __free_slab(s, page);
+}
+
+static void discard_slab(struct kmem_cache *s, struct page *page)
+{
+    dec_slabs_node(s, page_to_nid(page), page->objects);
+    free_slab(s, page);
+}
+
+/*
+ * Management of partially allocated slabs.
+ *
+ * list_lock must be held.
+ */
+static inline void add_partial(struct kmem_cache_node *n,
+                struct page *page, int tail)
+{
+    n->nr_partial++;
+    if (tail == DEACTIVATE_TO_TAIL)
+        list_add_tail(&page->lru, &n->partial);
+    else
+        list_add(&page->lru, &n->partial);
+}
+
+/*
+ * list_lock must be held.
+ */
+static inline void remove_partial(struct kmem_cache_node *n,
+                    struct page *page)
+{
+    list_del(&page->lru);
+    n->nr_partial--;
+}
+
+/*
+ * Remove slab from the partial list, freeze it and
+ * return the pointer to the freelist.
+ *
+ * Returns a list of objects or NULL if it fails.
+ *
+ * Must hold list_lock since we modify the partial list.
+ */
+static inline void *acquire_slab(struct kmem_cache *s,
+        struct kmem_cache_node *n, struct page *page,
+        int mode, int *objects)
+{
+    void *freelist;
+    unsigned long counters;
+    struct page new;
+
+    /*
+     * Zap the freelist and set the frozen bit.
+     * The old freelist is the list of objects for the
+     * per cpu allocation list.
+     */
+    freelist = page->freelist;
+    counters = page->counters;
+    new.counters = counters;
+    *objects = new.objects - new.inuse;
+    if (mode) {
+        new.inuse = page->objects;
+        new.freelist = NULL;
+    } else {
+        new.freelist = freelist;
+    }
+
+    VM_BUG_ON(new.frozen);
+    new.frozen = 1;
+
+    if (!__cmpxchg_double_slab(s, page,
+            freelist, counters,
+            new.freelist, new.counters,
+            "acquire_slab"))
+        return NULL;
+
+    remove_partial(n, page);
+    WARN_ON(!freelist);
+    return freelist;
+}
+
+static void put_cpu_partial(struct kmem_cache *s, struct page *page, int drain);
+static inline bool pfmemalloc_match(struct page *page, gfp_t gfpflags);
+
+/*
+ * Try to allocate a partial slab from a specific node.
+ */
+static void *get_partial_node(struct kmem_cache *s, struct kmem_cache_node *n,
+                struct kmem_cache_cpu *c, gfp_t flags)
+{
+    struct page *page, *page2;
+    void *object = NULL;
+    int available = 0;
+    int objects;
+
+    /*
+     * Racy check. If we mistakenly see no partial slabs then we
+     * just allocate an empty slab. If we mistakenly try to get a
+     * partial slab and there is none available then get_partials()
+     * will return NULL.
+     */
+    if (!n || !n->nr_partial)
+        return NULL;
+
+    spin_lock(&n->list_lock);
+    list_for_each_entry_safe(page, page2, &n->partial, lru) {
+        void *t;
+
+        if (!pfmemalloc_match(page, flags))
+            continue;
+
+        t = acquire_slab(s, n, page, object == NULL, &objects);
+        if (!t)
+            break;
+
+        available += objects;
+        if (!object) {
+            c->page = page;
+            stat(s, ALLOC_FROM_PARTIAL);
+            object = t;
+        } else {
+            put_cpu_partial(s, page, 0);
+            stat(s, CPU_PARTIAL_NODE);
+        }
+        if (kmem_cache_debug(s) || available > s->cpu_partial / 2)
+            break;
+
+    }
+    spin_unlock(&n->list_lock);
+    return object;
+}
+
+/*
+ * Get a page from somewhere. Search in increasing NUMA distances.
+ */
+static void *get_any_partial(struct kmem_cache *s, gfp_t flags,
+        struct kmem_cache_cpu *c)
+{
+#ifdef CONFIG_NUMA
+    struct zonelist *zonelist;
+    struct zoneref *z;
+    struct zone *zone;
+    enum zone_type high_zoneidx = gfp_zone(flags);
+    void *object;
+    unsigned int cpuset_mems_cookie;
+
+    /*
+     * The defrag ratio allows a configuration of the tradeoffs between
+     * inter node defragmentation and node local allocations. A lower
+     * defrag_ratio increases the tendency to do local allocations
+     * instead of attempting to obtain partial slabs from other nodes.
+     *
+     * If the defrag_ratio is set to 0 then kmalloc() always
+     * returns node local objects. If the ratio is higher then kmalloc()
+     * may return off node objects because partial slabs are obtained
+     * from other nodes and filled up.
+     *
+     * If /sys/kernel/slab/xx/defrag_ratio is set to 100 (which makes
+     * defrag_ratio = 1000) then every (well almost) allocation will
+     * first attempt to defrag slab caches on other nodes. This means
+     * scanning over all nodes to look for partial slabs which may be
+     * expensive if we do it every time we are trying to find a slab
+     * with available objects.
+     */
+    if (!s->remote_node_defrag_ratio ||
+            get_cycles() % 1024 > s->remote_node_defrag_ratio)
+        return NULL;
+
+    do {
+        cpuset_mems_cookie = get_mems_allowed();
+        zonelist = node_zonelist(slab_node(), flags);
+        for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
+            struct kmem_cache_node *n;
+
+            n = get_node(s, zone_to_nid(zone));
+
+            if (n && cpuset_zone_allowed_hardwall(zone, flags) &&
+                    n->nr_partial > s->min_partial) {
+                object = get_partial_node(s, n, c, flags);
+                if (object) {
+                    /*
+                     * Return the object even if
+                     * put_mems_allowed indicated that
+                     * the cpuset mems_allowed was
+                     * updated in parallel. It's a
+                     * harmless race between the alloc
+                     * and the cpuset update.
+                     */
+                    put_mems_allowed(cpuset_mems_cookie);
+                    return object;
+                }
+            }
+        }
+    } while (!put_mems_allowed(cpuset_mems_cookie));
+#endif
+    return NULL;
+}
+
+/*
+ * Get a partial page, lock it and return it.
+ */
+static void *get_partial(struct kmem_cache *s, gfp_t flags, int node,
+        struct kmem_cache_cpu *c)
+{
+    void *object;
+    int searchnode = (node == NUMA_NO_NODE) ? numa_node_id() : node;
+
+    object = get_partial_node(s, get_node(s, searchnode), c, flags);
+    if (object || node != NUMA_NO_NODE)
+        return object;
+
+    return get_any_partial(s, flags, c);
+}
+
+#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
+    stat(s, CMPXCHG_DOUBLE_CPU_FAIL);
+}
+
+static void init_kmem_cache_cpus(struct kmem_cache *s)
+{
+    int cpu;
+
+    for_each_possible_cpu(cpu)
+        per_cpu_ptr(s->cpu_slab, cpu)->tid = init_tid(cpu);
+}
+
+/*
+ * Remove the cpu slab
+ */
+static void deactivate_slab(struct kmem_cache *s, struct page *page, void *freelist)
+{
+    enum slab_modes { M_NONE, M_PARTIAL, M_FULL, M_FREE };
+    struct kmem_cache_node *n = get_node(s, page_to_nid(page));
+    int lock = 0;
+    enum slab_modes l = M_NONE, m = M_NONE;
+    void *nextfree;
+    int tail = DEACTIVATE_TO_HEAD;
+    struct page new;
+    struct page old;
+
+    if (page->freelist) {
+        stat(s, DEACTIVATE_REMOTE_FREES);
+        tail = DEACTIVATE_TO_TAIL;
+    }
+
+    /*
+     * Stage one: Free all available per cpu objects back
+     * to the page freelist while it is still frozen. Leave the
+     * last one.
+     *
+     * There is no need to take the list->lock because the page
+     * is still frozen.
+     */
+    while (freelist && (nextfree = get_freepointer(s, freelist))) {
+        void *prior;
+        unsigned long counters;
+
+        do {
+            prior = page->freelist;
+            counters = page->counters;
+            set_freepointer(s, freelist, prior);
+            new.counters = counters;
+            new.inuse--;
+            VM_BUG_ON(!new.frozen);
+
+        } while (!__cmpxchg_double_slab(s, page,
+            prior, counters,
+            freelist, new.counters,
+            "drain percpu freelist"));
+
+        freelist = nextfree;
+    }
+
+    /*
+     * Stage two: Ensure that the page is unfrozen while the
+     * list presence reflects the actual number of objects
+     * during unfreeze.
+     *
+     * We setup the list membership and then perform a cmpxchg
+     * with the count. If there is a mismatch then the page
+     * is not unfrozen but the page is on the wrong list.
+     *
+     * Then we restart the process which may have to remove
+     * the page from the list that we just put it on again
+     * because the number of objects in the slab may have
+     * changed.
+     */
+redo:
+
+    old.freelist = page->freelist;
+    old.counters = page->counters;
+    VM_BUG_ON(!old.frozen);
+
+    /* Determine target state of the slab */
+    new.counters = old.counters;
+    if (freelist) {
+        new.inuse--;
+        set_freepointer(s, freelist, old.freelist);
+        new.freelist = freelist;
+    } else
+        new.freelist = old.freelist;
+
+    new.frozen = 0;
+
+    if (!new.inuse && n->nr_partial > s->min_partial)
+        m = M_FREE;
+    else if (new.freelist) {
+        m = M_PARTIAL;
+        if (!lock) {
+            lock = 1;
+            /*
+             * Taking the spinlock removes the possiblity
+             * that acquire_slab() will see a slab page that
+             * is frozen
+             */
+            spin_lock(&n->list_lock);
+        }
+    } else {
+        m = M_FULL;
+        if (kmem_cache_debug(s) && !lock) {
+            lock = 1;
+            /*
+             * This also ensures that the scanning of full
+             * slabs from diagnostic functions will not see
+             * any frozen slabs.
+             */
+            spin_lock(&n->list_lock);
+        }
+    }
+
+    if (l != m) {
+
+        if (l == M_PARTIAL)
+
+            remove_partial(n, page);
+
+        else if (l == M_FULL)
+
+            remove_full(s, page);
+
+        if (m == M_PARTIAL) {
+
+            add_partial(n, page, tail);
+            stat(s, tail);
+
+        } else if (m == M_FULL) {
+
+            stat(s, DEACTIVATE_FULL);
+            add_full(s, n, page);
+
+        }
+    }
+
+    l = m;
+    if (!__cmpxchg_double_slab(s, page,
+                old.freelist, old.counters,
+                new.freelist, new.counters,
+                "unfreezing slab"))
+        goto redo;
+
+    if (lock)
+        spin_unlock(&n->list_lock);
+
+    if (m == M_FREE) {
+        stat(s, DEACTIVATE_EMPTY);
+        discard_slab(s, page);
+        stat(s, FREE_SLAB);
+    }
+}
+
+/*
+ * Unfreeze all the cpu partial slabs.
+ *
+ * This function must be called with interrupts disabled
+ * for the cpu using c (or some other guarantee must be there
+ * to guarantee no concurrent accesses).
+ */
+static void unfreeze_partials(struct kmem_cache *s,
+        struct kmem_cache_cpu *c)
+{
+    struct kmem_cache_node *n = NULL, *n2 = NULL;
+    struct page *page, *discard_page = NULL;
+
+    while ((page = c->partial)) {
+        struct page new;
+        struct page old;
+
+        c->partial = page->next;
+
+        n2 = get_node(s, page_to_nid(page));
+        if (n != n2) {
+            if (n)
+                spin_unlock(&n->list_lock);
+
+            n = n2;
+            spin_lock(&n->list_lock);
+        }
+
+        do {
+
+            old.freelist = page->freelist;
+            old.counters = page->counters;
+            VM_BUG_ON(!old.frozen);
+
+            new.counters = old.counters;
+            new.freelist = old.freelist;
+
+            new.frozen = 0;
+
+        } while (!__cmpxchg_double_slab(s, page,
+                old.freelist, old.counters,
+                new.freelist, new.counters,
+                "unfreezing slab"));
+
+        if (unlikely(!new.inuse && n->nr_partial > s->min_partial)) {
+            page->next = discard_page;
+            discard_page = page;
+        } else {
+            add_partial(n, page, DEACTIVATE_TO_TAIL);
+            stat(s, FREE_ADD_PARTIAL);
+        }
+    }
+
+    if (n)
+        spin_unlock(&n->list_lock);
+
+    while (discard_page) {
+        page = discard_page;
+        discard_page = discard_page->next;
+
+        stat(s, DEACTIVATE_EMPTY);
+        discard_slab(s, page);
+        stat(s, FREE_SLAB);
+    }
+}
+
+/*
+ * Put a page that was just frozen (in __slab_free) into a partial page
+ * slot if available. This is done without interrupts disabled and without
+ * preemption disabled. The cmpxchg is racy and may put the partial page
+ * onto a random cpus partial slot.
+ *
+ * If we did not find a slot then simply move all the partials to the
+ * per node partial list.
+ */
+static void put_cpu_partial(struct kmem_cache *s, struct page *page, int drain)
+{
+    struct page *oldpage;
+    int pages;
+    int pobjects;
+
+    do {
+        pages = 0;
+        pobjects = 0;
+        oldpage = this_cpu_read(s->cpu_slab->partial);
+
+        if (oldpage) {
+            pobjects = oldpage->pobjects;
+            pages = oldpage->pages;
+            if (drain && pobjects > s->cpu_partial) {
+                unsigned long flags;
+                /*
+                 * partial array is full. Move the existing
+                 * set to the per node partial list.
+                 */
+                local_irq_save(flags);
+                unfreeze_partials(s, this_cpu_ptr(s->cpu_slab));
+                local_irq_restore(flags);
+                oldpage = NULL;
+                pobjects = 0;
+                pages = 0;
+                stat(s, CPU_PARTIAL_DRAIN);
+            }
+        }
+
+        pages++;
+        pobjects += page->objects - page->inuse;
+
+        page->pages = pages;
+        page->pobjects = pobjects;
+        page->next = oldpage;
+
+    } while (this_cpu_cmpxchg(s->cpu_slab->partial, oldpage, page) != oldpage);
+}
+
+static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c)
+{
+    stat(s, CPUSLAB_FLUSH);
+    deactivate_slab(s, c->page, c->freelist);
+
+    c->tid = next_tid(c->tid);
+    c->page = NULL;
+    c->freelist = NULL;
+}
+
+/*
+ * Flush cpu slab.
+ *
+ * Called from IPI handler with interrupts disabled.
+ */
+static inline void __flush_cpu_slab(struct kmem_cache *s, int cpu)
+{
+    struct kmem_cache_cpu *c = per_cpu_ptr(s->cpu_slab, cpu);
+
+    if (likely(c)) {
+        if (c->page)
+            flush_slab(s, c);
+
+        unfreeze_partials(s, c);
+    }
+}
+
+static void flush_cpu_slab(void *d)
+{
+    struct kmem_cache *s = d;
+
+    __flush_cpu_slab(s, smp_processor_id());
+}
+
+static bool has_cpu_slab(int cpu, void *info)
+{
+    struct kmem_cache *s = info;
+    struct kmem_cache_cpu *c = per_cpu_ptr(s->cpu_slab, cpu);
+
+    return c->page || c->partial;
+}
+
+static void flush_all(struct kmem_cache *s)
+{
+    on_each_cpu_cond(has_cpu_slab, flush_cpu_slab, s, 1, GFP_ATOMIC);
+}
+
+/*
+ * Check if the objects in a per cpu structure fit numa
+ * locality expectations.
+ */
+static inline int node_match(struct page *page, int node)
+{
+#ifdef CONFIG_NUMA
+    if (!page || (node != NUMA_NO_NODE && page_to_nid(page) != node))
+        return 0;
+#endif
+    return 1;
+}
+
+static int count_free(struct page *page)
+{
+    return page->objects - page->inuse;
+}
+
+static unsigned long count_partial(struct kmem_cache_node *n,
+                    int (*get_count)(struct page *))
+{
+    unsigned long flags;
+    unsigned long x = 0;
+    struct page *page;
+
+    spin_lock_irqsave(&n->list_lock, flags);
+    list_for_each_entry(page, &n->partial, lru)
+        x += get_count(page);
+    spin_unlock_irqrestore(&n->list_lock, flags);
+    return x;
+}
+
+static inline unsigned long node_nr_objs(struct kmem_cache_node *n)
+{
+#ifdef CONFIG_SLEB_DEBUG
+    return atomic_long_read(&n->total_objects);
+#else
+    return 0;
+#endif
+}
+
+static noinline void
+slab_out_of_memory(struct kmem_cache *s, gfp_t gfpflags, int nid)
+{
+    int node;
+
+    printk(KERN_WARNING
+        "SLUB: Unable to allocate memory on node %d (gfp=0x%x)\n",
+        nid, gfpflags);
+    printk(KERN_WARNING "  cache: %s, object size: %d, buffer size: %d, "
+        "default order: %d, min order: %d\n", s->name, s->object_size,
+        s->size, oo_order(s->oo), oo_order(s->min));
+
+    if (oo_order(s->min) > get_order(s->object_size))
+        printk(KERN_WARNING "  %s debugging increased min order, use "
+               "slub_debug=O to disable.\n", s->name);
+
+    for_each_online_node(node) {
+        struct kmem_cache_node *n = get_node(s, node);
+        unsigned long nr_slabs;
+        unsigned long nr_objs;
+        unsigned long nr_free;
+
+        if (!n)
+            continue;
+
+        nr_free  = count_partial(n, count_free);
+        nr_slabs = node_nr_slabs(n);
+        nr_objs  = node_nr_objs(n);
+
+        printk(KERN_WARNING
+            "  node %d: slabs: %ld, objs: %ld, free: %ld\n",
+            node, nr_slabs, nr_objs, nr_free);
+    }
+}
+
+static inline void *new_slab_objects(struct kmem_cache *s, gfp_t flags,
+            int node, struct kmem_cache_cpu **pc)
+{
+    void *freelist;
+    struct kmem_cache_cpu *c = *pc;
+    struct page *page;
+
+    freelist = get_partial(s, flags, node, c);
+
+    if (freelist)
+        return freelist;
+
+    page = new_slab(s, flags, node);
+    if (page) {
+        c = __this_cpu_ptr(s->cpu_slab);
+        if (c->page)
+            flush_slab(s, c);
+
+        /*
+         * No other reference to the page yet so we can
+         * muck around with it freely without cmpxchg
+         */
+        freelist = page->freelist;
+        page->freelist = NULL;
+
+        stat(s, ALLOC_SLAB);
+        c->page = page;
+        *pc = c;
+    } else
+        freelist = NULL;
+
+    return freelist;
+}
+
+static inline bool pfmemalloc_match(struct page *page, gfp_t gfpflags)
+{
+    if (unlikely(PageSlabPfmemalloc(page)))
+        return gfp_pfmemalloc_allowed(gfpflags);
+
+    return true;
+}
+
+/*
+ * Check the page->freelist of a page and either transfer the freelist to the per cpu freelist
+ * or deactivate the page.
+ *
+ * The page is still frozen if the return value is not NULL.
+ *
+ * If this function returns NULL then the page has been unfrozen.
+ *
+ * This function must be called with interrupt disabled.
+ */
+static inline void *get_freelist(struct kmem_cache *s, struct page *page)
+{
+    struct page new;
+    unsigned long counters;
+    void *freelist;
+
+    do {
+        freelist = page->freelist;
+        counters = page->counters;
+
+        new.counters = counters;
+        VM_BUG_ON(!new.frozen);
+
+        new.inuse = page->objects;
+        new.frozen = freelist != NULL;
+
+    } while (!__cmpxchg_double_slab(s, page,
+        freelist, counters,
+        NULL, new.counters,
+        "get_freelist"));
+
+    return freelist;
+}
+
+/*
+ * Slow path. The lockless freelist is empty or we need to perform
+ * debugging duties.
+ *
+ * Processing is still very fast if new objects have been freed to the
+ * regular freelist. In that case we simply take over the regular freelist
+ * as the lockless freelist and zap the regular freelist.
+ *
+ * If that is not working then we fall back to the partial lists. We take the
+ * first element of the freelist as the object to allocate now and move the
+ * rest of the freelist to the lockless freelist.
+ *
+ * And if we were unable to get a new slab from the partial slab lists then
+ * we need to allocate a new slab. This is the slowest path since it involves
+ * a call to the page allocator and the setup of a new slab.
+ */
+static void *__slab_alloc(struct kmem_cache *s, gfp_t gfpflags, int node,
+              unsigned long addr, struct kmem_cache_cpu *c)
+{
+    void *freelist;
+    struct page *page;
+    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
+
+    page = c->page;
+    if (!page)
+        goto new_slab;
+redo:
+
+    if (unlikely(!node_match(page, node))) {
+        stat(s, ALLOC_NODE_MISMATCH);
+        deactivate_slab(s, page, c->freelist);
+        c->page = NULL;
+        c->freelist = NULL;
+        goto new_slab;
+    }
+
+    /*
+     * By rights, we should be searching for a slab page that was
+     * PFMEMALLOC but right now, we are losing the pfmemalloc
+     * information when the page leaves the per-cpu allocator
+     */
+    if (unlikely(!pfmemalloc_match(page, gfpflags))) {
+        deactivate_slab(s, page, c->freelist);
+        c->page = NULL;
+        c->freelist = NULL;
+        goto new_slab;
+    }
+
+    /* must check again c->freelist in case of cpu migration or IRQ */
+    freelist = c->freelist;
+    if (freelist)
+        goto load_freelist;
+
+    stat(s, ALLOC_SLOWPATH);
+
+    freelist = get_freelist(s, page);
+
+    if (!freelist) {
+        c->page = NULL;
+        stat(s, DEACTIVATE_BYPASS);
+        goto new_slab;
+    }
+
+    stat(s, ALLOC_REFILL);
+
+load_freelist:
+    /*
+     * freelist is pointing to the list of objects to be used.
+     * page is pointing to the page from which the objects are obtained.
+     * That page must be frozen for per cpu allocations to work.
+     */
+    VM_BUG_ON(!c->page->frozen);
+    c->freelist = get_freepointer(s, freelist);
+    c->tid = next_tid(c->tid);
+    local_irq_restore(flags);
+    return freelist;
+
+new_slab:
+
+    if (c->partial) {
+        page = c->page = c->partial;
+        c->partial = page->next;
+        stat(s, CPU_PARTIAL_ALLOC);
+        c->freelist = NULL;
+        goto redo;
+    }
+
+    freelist = new_slab_objects(s, gfpflags, node, &c);
+
+    if (unlikely(!freelist)) {
+        if (!(gfpflags & __GFP_NOWARN) && printk_ratelimit())
+            slab_out_of_memory(s, gfpflags, node);
+
+        local_irq_restore(flags);
+        return NULL;
+    }
+
+    page = c->page;
+    if (likely(!kmem_cache_debug(s) && pfmemalloc_match(page, gfpflags)))
+        goto load_freelist;
+
+    /* Only entered in the debug case */
+    if (kmem_cache_debug(s) && !alloc_debug_processing(s, page, freelist, addr))
+        goto new_slab;    /* Slab failed checks. Next slab needed */
+
+    deactivate_slab(s, page, get_freepointer(s, freelist));
+    c->page = NULL;
+    c->freelist = NULL;
+    local_irq_restore(flags);
+    return freelist;
+}
+
+/*
+ * Inlined fastpath so that allocation functions (kmalloc, kmem_cache_alloc)
+ * have the fastpath folded into their functions. So no function call
+ * overhead for requests that can be satisfied on the fastpath.
+ *
+ * The fastpath works by first checking if the lockless freelist can be used.
+ * If not then __slab_alloc is called for slow processing.
+ *
+ * Otherwise we can simply pick the next object from the lockless free list.
+ */
+static __always_inline void *slab_alloc_node(struct kmem_cache *s,
+        gfp_t gfpflags, int node, unsigned long addr)
+{
+    void **object;
+    struct kmem_cache_cpu *c;
+    struct page *page;
+    unsigned long tid;
+
+    if (slab_pre_alloc_hook(s, gfpflags))
+        return NULL;
+
+    s = memcg_kmem_get_cache(s, gfpflags);
+redo:
+    /*
+     * 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.
+     *
+     * Preemption is disabled for the retrieval of the tid because that
+     * must occur from the current processor. We cannot allow rescheduling
+     * on a different processor between the determination of the pointer
+     * and the retrieval of the tid.
+     */
+    preempt_disable();
+    c = __this_cpu_ptr(s->cpu_slab);
+
+    /*
+     * 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;
+    preempt_enable();
+
+    object = c->freelist;
+    page = c->page;
+    if (unlikely(!object || !node_match(page, node)))
+        object = __slab_alloc(s, gfpflags, node, addr, c);
+
+    else {
+        void *next_object = get_freepointer_safe(s, object);
+
+        /*
+         * The cmpxchg will only match if there was no additional
+         * 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,
+                next_object, next_tid(tid)))) {
+
+            note_cmpxchg_failure("slab_alloc", s, tid);
+            goto redo;
+        }
+        prefetch_freepointer(s, next_object);
+        stat(s, ALLOC_FASTPATH);
+    }
+
+    if (unlikely(gfpflags & __GFP_ZERO) && object)
+        memset(object, 0, s->object_size);
+
+    slab_post_alloc_hook(s, gfpflags, object);
+
+    return object;
+}
+
+static __always_inline void *slab_alloc(struct kmem_cache *s,
+        gfp_t gfpflags, unsigned long addr)
+{
+    return slab_alloc_node(s, gfpflags, NUMA_NO_NODE, addr);
+}
+
+void *kmem_cache_alloc(struct kmem_cache *s, gfp_t gfpflags)
+{
+    void *ret = slab_alloc(s, gfpflags, _RET_IP_);
+
+    trace_kmem_cache_alloc(_RET_IP_, ret, s->object_size, s->size, gfpflags);
+
+    return ret;
+}
+EXPORT_SYMBOL(kmem_cache_alloc);
+
+#ifdef CONFIG_TRACING
+void *kmem_cache_alloc_trace(struct kmem_cache *s, gfp_t gfpflags, size_t size)
+{
+    void *ret = slab_alloc(s, gfpflags, _RET_IP_);
+    trace_kmalloc(_RET_IP_, ret, size, s->size, gfpflags);
+    return ret;
+}
+EXPORT_SYMBOL(kmem_cache_alloc_trace);
+
+void *kmalloc_order_trace(size_t size, gfp_t flags, unsigned int order)
+{
+    void *ret = kmalloc_order(size, flags, order);
+    trace_kmalloc(_RET_IP_, ret, size, PAGE_SIZE << order, flags);
+    return ret;
+}
+EXPORT_SYMBOL(kmalloc_order_trace);
+#endif
+
+#ifdef CONFIG_NUMA
+void *kmem_cache_alloc_node(struct kmem_cache *s, gfp_t gfpflags, int node)
+{
+    void *ret = slab_alloc_node(s, gfpflags, node, _RET_IP_);
+
+    trace_kmem_cache_alloc_node(_RET_IP_, ret,
+                    s->object_size, s->size, gfpflags, node);
+
+    return ret;
+}
+EXPORT_SYMBOL(kmem_cache_alloc_node);
+
+#ifdef CONFIG_TRACING
+void *kmem_cache_alloc_node_trace(struct kmem_cache *s,
+                    gfp_t gfpflags,
+                    int node, size_t size)
+{
+    void *ret = slab_alloc_node(s, gfpflags, node, _RET_IP_);
+
+    trace_kmalloc_node(_RET_IP_, ret,
+               size, s->size, gfpflags, node);
+    return ret;
+}
+EXPORT_SYMBOL(kmem_cache_alloc_node_trace);
+#endif
+#endif
+
+/*
+ * Slow patch handling. This may still be called frequently since objects
+ * have a longer lifetime than the cpu slabs in most processing loads.
+ *
+ * So we still attempt to reduce cache line usage. Just take the slab
+ * lock and free the item. If there is no additional partial page
+ * handling required then we can return immediately.
+ */
+static void __slab_free(struct kmem_cache *s, struct page *page,
+            void *x, unsigned long addr)
+{
+    void *prior;
+    void **object = (void *)x;
+    int was_frozen;
+    struct page new;
+    unsigned long counters;
+    struct kmem_cache_node *n = NULL;
+    unsigned long uninitialized_var(flags);
+
+    stat(s, FREE_SLOWPATH);
+
+    if (kmem_cache_debug(s) &&
+        !(n = free_debug_processing(s, page, x, addr, &flags)))
+        return;
+
+    do {
+        if (unlikely(n)) {
+            spin_unlock_irqrestore(&n->list_lock, flags);
+            n = NULL;
+        }
+        prior = page->freelist;
+        counters = page->counters;
+        set_freepointer(s, object, prior);
+        new.counters = counters;
+        was_frozen = new.frozen;
+        new.inuse--;
+        if ((!new.inuse || !prior) && !was_frozen) {
+
+            if (!kmem_cache_debug(s) && !prior)
+
+                /*
+                 * Slab was on no list before and will be partially empty
+                 * We can defer the list move and instead freeze it.
+                 */
+                new.frozen = 1;
+
+            else { /* Needs to be taken off a list */
+
+                            n = get_node(s, page_to_nid(page));
+                /*
+                 * Speculatively acquire the list_lock.
+                 * If the cmpxchg does not succeed then we may
+                 * drop the list_lock without any processing.
+                 *
+                 * Otherwise the list_lock will synchronize with
+                 * other processors updating the list of slabs.
+                 */
+                spin_lock_irqsave(&n->list_lock, flags);
+
+            }
+        }
+
+    } while (!cmpxchg_double_slab(s, page,
+        prior, counters,
+        object, new.counters,
+        "__slab_free"));
+
+    if (likely(!n)) {
+
+        /*
+         * If we just froze the page then put it onto the
+         * per cpu partial list.
+         */
+        if (new.frozen && !was_frozen) {
+            put_cpu_partial(s, page, 1);
+            stat(s, CPU_PARTIAL_FREE);
+        }
+        /*
+         * The list lock was not taken therefore no list
+         * activity can be necessary.
+         */
+                if (was_frozen)
+                        stat(s, FREE_FROZEN);
+                return;
+        }
+
+    if (unlikely(!new.inuse && n->nr_partial > s->min_partial))
+        goto slab_empty;
+
+    /*
+     * 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, page);
+        add_partial(n, page, DEACTIVATE_TO_TAIL);
+        stat(s, FREE_ADD_PARTIAL);
+    }
+    spin_unlock_irqrestore(&n->list_lock, flags);
+    return;
+
+slab_empty:
+    if (prior) {
+        /*
+         * Slab on the partial list.
+         */
+        remove_partial(n, page);
+        stat(s, FREE_REMOVE_PARTIAL);
+    } else
+        /* Slab must be on the full list */
+        remove_full(s, page);
+
+    spin_unlock_irqrestore(&n->list_lock, flags);
+    stat(s, FREE_SLAB);
+    discard_slab(s, page);
+}
+
+/*
+ * Fastpath with forced inlining to produce a kfree and kmem_cache_free that
+ * can perform fastpath freeing without additional function calls.
+ *
+ * The fastpath is only possible if we are freeing to the current cpu slab
+ * of this processor. This typically the case if we have just allocated
+ * the item before.
+ *
+ * If fastpath is not possible then fall back to __slab_free where we deal
+ * with all sorts of special processing.
+ */
+static __always_inline void slab_free(struct kmem_cache *s,
+            struct page *page, void *x, unsigned long addr)
+{
+    void **object = (void *)x;
+    struct kmem_cache_cpu *c;
+    unsigned long tid;
+
+    slab_free_hook(s, x);
+
+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.
+     */
+    preempt_disable();
+    c = __this_cpu_ptr(s->cpu_slab);
+
+    tid = c->tid;
+    preempt_enable();
+
+    if (likely(page == c->page)) {
+        set_freepointer(s, object, c->freelist);
+
+        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;
+        }
+        stat(s, FREE_FASTPATH);
+    } else
+        __slab_free(s, page, x, addr);
+
+}
+
+void kmem_cache_free(struct kmem_cache *s, void *x)
+{
+    s = cache_from_obj(s, x);
+    if (!s)
+        return;
+    slab_free(s, virt_to_head_page(x), x, _RET_IP_);
+    trace_kmem_cache_free(_RET_IP_, x);
+}
+EXPORT_SYMBOL(kmem_cache_free);
+
+/*
+ * Object placement in a slab is made very easy because we always start at
+ * offset 0. If we tune the size of the object to the alignment then we can
+ * get the required alignment by putting one properly sized object after
+ * another.
+ *
+ * Notice that the allocation order determines the sizes of the per cpu
+ * caches. Each processor has always one slab available for allocations.
+ * Increasing the allocation order reduces the number of times that slabs
+ * must be moved on and off the partial lists and is therefore a factor in
+ * locking overhead.
+ */
+
+/*
+ * Mininum / Maximum order of slab pages. This influences locking overhead
+ * and slab fragmentation. A higher order reduces the number of partial slabs
+ * and increases the number of allocations possible without having to
+ * take the list_lock.
+ */
+static int slub_min_order;
+static int slub_max_order = PAGE_ALLOC_COSTLY_ORDER;
+static int slub_min_objects;
+
+/*
+ * Merge control. If this is set then no merging of slab caches will occur.
+ * (Could be removed. This was introduced to pacify the merge skeptics.)
+ */
+static int slub_nomerge;
+
+/*
+ * Calculate the order of allocation given an slab object size.
+ *
+ * The order of allocation has significant impact on performance and other
+ * system components. Generally order 0 allocations should be preferred since
+ * order 0 does not cause fragmentation in the page allocator. Larger objects
+ * be problematic to put into order 0 slabs because there may be too much
+ * unused space left. We go to a higher order if more than 1/16th of the slab
+ * would be wasted.
+ *
+ * In order to reach satisfactory performance we must ensure that a minimum
+ * number of objects is in one slab. Otherwise we may generate too much
+ * activity on the partial lists which requires taking the list_lock. This is
+ * less a concern for large slabs though which are rarely used.
+ *
+ * slub_max_order specifies the order where we begin to stop considering the
+ * number of objects in a slab as critical. If we reach slub_max_order then
+ * we try to keep the page order as low as possible. So we accept more waste
+ * of space in favor of a small page order.
+ *
+ * Higher order allocations also allow the placement of more objects in a
+ * slab and thereby reduce object handling overhead. If the user has
+ * requested a higher mininum order then we start with that one instead of
+ * the smallest order which will fit the object.
+ */
+static inline int slab_order(int size, int min_objects,
+                int max_order, int fract_leftover, int reserved)
+{
+    int order;
+    int rem;
+    int min_order = slub_min_order;
+
+    if (order_objects(min_order, size, reserved) > MAX_OBJS_PER_PAGE)
+        return get_order(size * MAX_OBJS_PER_PAGE) - 1;
+
+    for (order = max(min_order,
+                fls(min_objects * size - 1) - PAGE_SHIFT);
+            order <= max_order; order++) {
+
+        unsigned long slab_size = PAGE_SIZE << order;
+
+        if (slab_size < min_objects * size + reserved)
+            continue;
+
+        rem = (slab_size - reserved) % size;
+
+        if (rem <= slab_size / fract_leftover)
+            break;
+
+    }
+
+    return order;
+}
+
+static inline int calculate_order(int size, int reserved)
+{
+    int order;
+    int min_objects;
+    int fraction;
+    int max_objects;
+
+    /*
+     * Attempt to find best configuration for a slab. This
+     * works by first attempting to generate a layout with
+     * the best configuration and backing off gradually.
+     *
+     * First we reduce the acceptable waste in a slab. Then
+     * we reduce the minimum objects required in a slab.
+     */
+    min_objects = slub_min_objects;
+    if (!min_objects)
+        min_objects = 4 * (fls(nr_cpu_ids) + 1);
+    max_objects = order_objects(slub_max_order, size, reserved);
+    min_objects = min(min_objects, max_objects);
+
+    while (min_objects > 1) {
+        fraction = 16;
+        while (fraction >= 4) {
+            order = slab_order(size, min_objects,
+                    slub_max_order, fraction, reserved);
+            if (order <= slub_max_order)
+                return order;
+            fraction /= 2;
+        }
+        min_objects--;
+    }
+
+    /*
+     * We were unable to place multiple objects in a slab. Now
+     * lets see if we can place a single object there.
+     */
+    order = slab_order(size, 1, slub_max_order, 1, reserved);
+    if (order <= slub_max_order)
+        return order;
+
+    /*
+     * Doh this slab cannot be placed using slub_max_order.
+     */
+    order = slab_order(size, 1, MAX_ORDER, 1, reserved);
+    if (order < MAX_ORDER)
+        return order;
+    return -ENOSYS;
+}
+
+static void
+init_kmem_cache_node(struct kmem_cache_node *n)
+{
+    n->nr_partial = 0;
+    spin_lock_init(&n->list_lock);
+    INIT_LIST_HEAD(&n->partial);
+#ifdef CONFIG_SLEB_DEBUG
+    atomic_long_set(&n->nr_slabs, 0);
+    atomic_long_set(&n->total_objects, 0);
+    INIT_LIST_HEAD(&n->full);
+#endif
+}
+
+static inline int alloc_kmem_cache_cpus(struct kmem_cache *s)
+{
+    BUILD_BUG_ON(PERCPU_DYNAMIC_EARLY_SIZE <
+            KMALLOC_SHIFT_HIGH * sizeof(struct kmem_cache_cpu));
+
+    /*
+     * Must align to double word boundary for the double cmpxchg
+     * instructions to work; see __pcpu_double_call_return_bool().
+     */
+    s->cpu_slab = __alloc_percpu(sizeof(struct kmem_cache_cpu),
+                     2 * sizeof(void *));
+
+    if (!s->cpu_slab)
+        return 0;
+
+    init_kmem_cache_cpus(s);
+
+    return 1;
+}
+
+static struct kmem_cache *kmem_cache_node;
+
+/*
+ * No kmalloc_node yet so do it by hand. We know that this is the first
+ * slab on the node for this slabcache. There are no concurrent accesses
+ * possible.
+ *
+ * Note that this function only works on the kmalloc_node_cache
+ * when allocating for the kmalloc_node_cache. This is used for bootstrapping
+ * memory on a fresh node that has no slab structures yet.
+ */
+static void early_kmem_cache_node_alloc(int node)
+{
+    struct page *page;
+    struct kmem_cache_node *n;
+
+    BUG_ON(kmem_cache_node->size < sizeof(struct kmem_cache_node));
+
+    page = new_slab(kmem_cache_node, GFP_NOWAIT, node);
+
+    BUG_ON(!page);
+    if (page_to_nid(page) != node) {
+        printk(KERN_ERR "SLUB: Unable to allocate memory from "
+                "node %d\n", node);
+        printk(KERN_ERR "SLUB: Allocating a useless per node structure "
+                "in order to be able to continue\n");
+    }
+
+    n = page->freelist;
+    BUG_ON(!n);
+    page->freelist = get_freepointer(kmem_cache_node, n);
+    page->inuse = 1;
+    page->frozen = 0;
+    kmem_cache_node->node[node] = n;
+#ifdef CONFIG_SLEB_DEBUG
+    init_object(kmem_cache_node, n, SLUB_RED_ACTIVE);
+    init_tracking(kmem_cache_node, n);
+#endif
+    init_kmem_cache_node(n);
+    inc_slabs_node(kmem_cache_node, node, page->objects);
+
+    add_partial(n, page, DEACTIVATE_TO_HEAD);
+}
+
+static void free_kmem_cache_nodes(struct kmem_cache *s)
+{
+    int node;
+
+    for_each_node_state(node, N_NORMAL_MEMORY) {
+        struct kmem_cache_node *n = s->node[node];
+
+        if (n)
+            kmem_cache_free(kmem_cache_node, n);
+
+        s->node[node] = NULL;
+    }
+}
+
+static int init_kmem_cache_nodes(struct kmem_cache *s)
+{
+    int node;
+
+    for_each_node_state(node, N_NORMAL_MEMORY) {
+        struct kmem_cache_node *n;
+
+        if (slab_state == DOWN) {
+            early_kmem_cache_node_alloc(node);
+            continue;
+        }
+        n = kmem_cache_alloc_node(kmem_cache_node,
+                        GFP_KERNEL, node);
+
+        if (!n) {
+            free_kmem_cache_nodes(s);
+            return 0;
+        }
+
+        s->node[node] = n;
+        init_kmem_cache_node(n);
+    }
+    return 1;
+}
+
+static void set_min_partial(struct kmem_cache *s, unsigned long min)
+{
+    if (min < MIN_PARTIAL)
+        min = MIN_PARTIAL;
+    else if (min > MAX_PARTIAL)
+        min = MAX_PARTIAL;
+    s->min_partial = min;
+}
+
+/*
+ * calculate_sizes() determines the order and the distribution of data within
+ * a slab object.
+ */
+static int calculate_sizes(struct kmem_cache *s, int forced_order)
+{
+    unsigned long flags = s->flags;
+    unsigned long size = s->object_size;
+    int order;
+
+    /*
+     * Round up object size to the next word boundary. We can only
+     * place the free pointer at word boundaries and this determines
+     * the possible location of the free pointer.
+     */
+    size = ALIGN(size, sizeof(void *));
+
+#ifdef CONFIG_SLEB_DEBUG
+    /*
+     * Determine if we can poison the object itself. If the user of
+     * the slab may touch the object after free or before allocation
+     * then we should never poison the object itself.
+     */
+    if ((flags & SLAB_POISON) && !(flags & SLAB_DESTROY_BY_RCU) &&
+            !s->ctor)
+        s->flags |= __OBJECT_POISON;
+    else
+        s->flags &= ~__OBJECT_POISON;
+
+
+    /*
+     * If we are Redzoning then check if there is some space between the
+     * end of the object and the free pointer. If not then add an
+     * additional word to have some bytes to store Redzone information.
+     */
+    if ((flags & SLAB_RED_ZONE) && size == s->object_size)
+        size += sizeof(void *);
+#endif
+
+    /*
+     * With that we have determined the number of bytes in actual use
+     * by the object. This is the potential offset to the free pointer.
+     */
+    s->inuse = size;
+
+    if (((flags & (SLAB_DESTROY_BY_RCU | SLAB_POISON)) ||
+        s->ctor)) {
+        /*
+         * Relocate free pointer after the object if it is not
+         * permitted to overwrite the first word of the object on
+         * kmem_cache_free.
+         *
+         * This is the case if we do RCU, have a constructor or
+         * destructor or are poisoning the objects.
+         */
+        s->offset = size;
+        size += sizeof(void *);
+    }
+
+#ifdef CONFIG_SLEB_DEBUG
+    if (flags & SLAB_STORE_USER)
+        /*
+         * Need to store information about allocs and frees after
+         * the object.
+         */
+        size += 2 * sizeof(struct track);
+
+    if (flags & SLAB_RED_ZONE)
+        /*
+         * Add some empty padding so that we can catch
+         * overwrites from earlier objects rather than let
+         * tracking information or the free pointer be
+         * corrupted if a user writes before the start
+         * of the object.
+         */
+        size += sizeof(void *);
+#endif
+
+    /*
+     * SLUB stores one object immediately after another beginning from
+     * offset 0. In order to align the objects we have to simply size
+     * each object to conform to the alignment.
+     */
+    size = ALIGN(size, s->align);
+    s->size = size;
+    if (forced_order >= 0)
+        order = forced_order;
+    else
+        order = calculate_order(size, s->reserved);
+
+    if (order < 0)
+        return 0;
+
+    s->allocflags = 0;
+    if (order)
+        s->allocflags |= __GFP_COMP;
+
+    if (s->flags & SLAB_CACHE_DMA)
+        s->allocflags |= GFP_DMA;
+
+    if (s->flags & SLAB_RECLAIM_ACCOUNT)
+        s->allocflags |= __GFP_RECLAIMABLE;
+
+    /*
+     * Determine the number of objects per slab
+     */
+    s->oo = oo_make(order, size, s->reserved);
+    s->min = oo_make(get_order(size), size, s->reserved);
+    if (oo_objects(s->oo) > oo_objects(s->max))
+        s->max = s->oo;
+
+    return !!oo_objects(s->oo);
+}
+
+static int kmem_cache_open(struct kmem_cache *s, unsigned long flags)
+{
+    s->flags = kmem_cache_flags(s->size, flags, s->name, s->ctor);
+    s->reserved = 0;
+
+    if (need_reserve_slab_rcu && (s->flags & SLAB_DESTROY_BY_RCU))
+        s->reserved = sizeof(struct rcu_head);
+
+    if (!calculate_sizes(s, -1))
+        goto error;
+    if (disable_higher_order_debug) {
+        /*
+         * Disable debugging flags that store metadata if the min slab
+         * order increased.
+         */
+        if (get_order(s->size) > get_order(s->object_size)) {
+            s->flags &= ~DEBUG_METADATA_FLAGS;
+            s->offset = 0;
+            if (!calculate_sizes(s, -1))
+                goto error;
+        }
+    }
+
+#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
+    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
+    if (system_has_cmpxchg_double() && (s->flags & SLAB_DEBUG_FLAGS) == 0)
+        /* Enable fast mode */
+        s->flags |= __CMPXCHG_DOUBLE;
+#endif
+
+    /*
+     * The larger the object size is, the more pages we want on the partial
+     * list to avoid pounding the page allocator excessively.
+     */
+    set_min_partial(s, ilog2(s->size) / 2);
+
+    /*
+     * cpu_partial determined the maximum number of objects kept in the
+     * per cpu partial lists of a processor.
+     *
+     * Per cpu partial lists mainly contain slabs that just have one
+     * object freed. If they are used for allocation then they can be
+     * filled up again with minimal effort. The slab will never hit the
+     * per node partial lists and therefore no locking will be required.
+     *
+     * This setting also determines
+     *
+     * A) The number of objects from per cpu partial slabs dumped to the
+     *    per node list when we reach the limit.
+     * B) The number of objects in cpu partial slabs to extract from the
+     *    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))
+        s->cpu_partial = 0;
+    else if (s->size >= PAGE_SIZE)
+        s->cpu_partial = 2;
+    else if (s->size >= 1024)
+        s->cpu_partial = 6;
+    else if (s->size >= 256)
+        s->cpu_partial = 13;
+    else
+        s->cpu_partial = 30;
+
+#ifdef CONFIG_NUMA
+    s->remote_node_defrag_ratio = 1000;
+#endif
+    if (!init_kmem_cache_nodes(s))
+        goto error;
+
+    if (alloc_kmem_cache_cpus(s))
+        return 0;
+
+    free_kmem_cache_nodes(s);
+error:
+    if (flags & SLAB_PANIC)
+        panic("Cannot create slab %s size=%lu realsize=%u "
+            "order=%u offset=%u flags=%lx\n",
+            s->name, (unsigned long)s->size, s->size, oo_order(s->oo),
+            s->offset, flags);
+    return -EINVAL;
+}
+
+static void list_slab_objects(struct kmem_cache *s, struct page *page,
+                            const char *text)
+{
+#ifdef CONFIG_SLEB_DEBUG
+    void *addr = page_address(page);
+    void *p;
+    unsigned long *map = kzalloc(BITS_TO_LONGS(page->objects) *
+                     sizeof(long), GFP_ATOMIC);
+    if (!map)
+        return;
+    slab_err(s, page, text, s->name);
+    slab_lock(page);
+
+    get_map(s, page, map);
+    for_each_object(p, s, addr, page->objects) {
+
+        if (!test_bit(slab_index(p, s, addr), map)) {
+            printk(KERN_ERR "INFO: Object 0x%p @offset=%tu\n",
+                            p, p - addr);
+            print_tracking(s, p);
+        }
+    }
+    slab_unlock(page);
+    kfree(map);
+#endif
+}
+
+/*
+ * Attempt to free all partial slabs on a node.
+ * This is called from kmem_cache_close(). We must be the last thread
+ * using the cache and therefore we do not need to lock anymore.
+ */
+static void free_partial(struct kmem_cache *s, struct kmem_cache_node *n)
+{
+    struct page *page, *h;
+
+    list_for_each_entry_safe(page, h, &n->partial, lru) {
+        if (!page->inuse) {
+            remove_partial(n, page);
+            discard_slab(s, page);
+        } else {
+            list_slab_objects(s, page,
+            "Objects remaining in %s on kmem_cache_close()");
+        }
+    }
+}
+
+/*
+ * Release all resources used by a slab cache.
+ */
+static inline int kmem_cache_close(struct kmem_cache *s)
+{
+    int node;
+
+    flush_all(s);
+    /* Attempt to free all objects */
+    for_each_node_state(node, N_NORMAL_MEMORY) {
+        struct kmem_cache_node *n = get_node(s, node);
+
+        free_partial(s, n);
+        if (n->nr_partial || slabs_node(s, node))
+            return 1;
+    }
+    free_percpu(s->cpu_slab);
+    free_kmem_cache_nodes(s);
+    return 0;
+}
+
+int __kmem_cache_shutdown(struct kmem_cache *s)
+{
+    int rc = kmem_cache_close(s);
+
+    if (!rc) {
+        /*
+         * We do the same lock strategy around sysfs_slab_add, see
+         * __kmem_cache_create. Because this is pretty much the last
+         * operation we do and the lock will be released shortly after
+         * that in slab_common.c, we could just move sysfs_slab_remove
+         * to a later point in common code. We should do that when we
+         * have a common sysfs framework for all allocators.
+         */
+        mutex_unlock(&slab_mutex);
+        sysfs_slab_remove(s);
+        mutex_lock(&slab_mutex);
+    }
+
+    return rc;
+}
+
+/********************************************************************
+ *        Kmalloc subsystem
+ *******************************************************************/
+
+static int __init setup_slub_min_order(char *str)
+{
+    get_option(&str, &slub_min_order);
+
+    return 1;
+}
+
+__setup("slub_min_order=", setup_slub_min_order);
+
+static int __init setup_slub_max_order(char *str)
+{
+    get_option(&str, &slub_max_order);
+    slub_max_order = min(slub_max_order, MAX_ORDER - 1);
+
+    return 1;
+}
+
+__setup("slub_max_order=", setup_slub_max_order);
+
+static int __init setup_slub_min_objects(char *str)
+{
+    get_option(&str, &slub_min_objects);
+
+    return 1;
+}
+
+__setup("slub_min_objects=", setup_slub_min_objects);
+
+static int __init setup_slub_nomerge(char *str)
+{
+    slub_nomerge = 1;
+    return 1;
+}
+
+__setup("slub_nomerge", setup_slub_nomerge);
+
+void *__kmalloc(size_t size, gfp_t flags)
+{
+    struct kmem_cache *s;
+    void *ret;
+
+    if (unlikely(size > KMALLOC_MAX_CACHE_SIZE))
+        return kmalloc_large(size, flags);
+
+    s = kmalloc_slab(size, flags);
+
+    if (unlikely(ZERO_OR_NULL_PTR(s)))
+        return s;
+
+    ret = slab_alloc(s, flags, _RET_IP_);
+
+    trace_kmalloc(_RET_IP_, ret, size, s->size, flags);
+
+    return ret;
+}
+EXPORT_SYMBOL(__kmalloc);
+
+#ifdef CONFIG_NUMA
+static void *kmalloc_large_node(size_t size, gfp_t flags, int node)
+{
+    struct page *page;
+    void *ptr = NULL;
+
+    flags |= __GFP_COMP | __GFP_NOTRACK | __GFP_KMEMCG;
+    page = alloc_pages_node(node, flags, get_order(size));
+    if (page)
+        ptr = page_address(page);
+
+    kmemleak_alloc(ptr, size, 1, flags);
+    return ptr;
+}
+
+void *__kmalloc_node(size_t size, gfp_t flags, int node)
+{
+    struct kmem_cache *s;
+    void *ret;
+
+    if (unlikely(size > KMALLOC_MAX_CACHE_SIZE)) {
+        ret = kmalloc_large_node(size, flags, node);
+
+        trace_kmalloc_node(_RET_IP_, ret,
+                   size, PAGE_SIZE << get_order(size),
+                   flags, node);
+
+        return ret;
+    }
+
+    s = kmalloc_slab(size, flags);
+
+    if (unlikely(ZERO_OR_NULL_PTR(s)))
+        return s;
+
+    ret = slab_alloc_node(s, flags, node, _RET_IP_);
+
+    trace_kmalloc_node(_RET_IP_, ret, size, s->size, flags, node);
+
+    return ret;
+}
+EXPORT_SYMBOL(__kmalloc_node);
+#endif
+
+size_t ksize(const void *object)
+{
+    struct page *page;
+
+    if (unlikely(object == ZERO_SIZE_PTR))
+        return 0;
+
+    page = virt_to_head_page(object);
+
+    if (unlikely(!PageSlab(page))) {
+        WARN_ON(!PageCompound(page));
+        return PAGE_SIZE << compound_order(page);
+    }
+
+    return slab_ksize(page->slab_cache);
+}
+EXPORT_SYMBOL(ksize);
+
+#ifdef CONFIG_SLEB_DEBUG
+bool verify_mem_not_deleted(const void *x)
+{
+    struct page *page;
+    void *object = (void *)x;
+    unsigned long flags;
+    bool rv;
+
+    if (unlikely(ZERO_OR_NULL_PTR(x)))
+        return false;
+
+    local_irq_save(flags);
+
+    page = virt_to_head_page(x);
+    if (unlikely(!PageSlab(page))) {
+        /* maybe it was from stack? */
+        rv = true;
+        goto out_unlock;
+    }
+
+    slab_lock(page);
+    if (on_freelist(page->slab_cache, page, object)) {
+        object_err(page->slab_cache, page, object, "Object is on free-list");
+        rv = false;
+    } else {
+        rv = true;
+    }
+    slab_unlock(page);
+
+out_unlock:
+    local_irq_restore(flags);
+    return rv;
+}
+EXPORT_SYMBOL(verify_mem_not_deleted);
+#endif
+
+void kfree(const void *x)
+{
+    struct page *page;
+    void *object = (void *)x;
+
+    trace_kfree(_RET_IP_, x);
+
+    if (unlikely(ZERO_OR_NULL_PTR(x)))
+        return;
+
+    page = virt_to_head_page(x);
+    if (unlikely(!PageSlab(page))) {
+        BUG_ON(!PageCompound(page));
+        kmemleak_free(x);
+        __free_memcg_kmem_pages(page, compound_order(page));
+        return;
+    }
+    slab_free(page->slab_cache, page, object, _RET_IP_);
+}
+EXPORT_SYMBOL(kfree);
+
+/*
+ * kmem_cache_shrink removes empty slabs from the partial lists and sorts
+ * the remaining slabs by the number of items in use. The slabs with the
+ * most items in use come first. New allocations will then fill those up
+ * and thus they can be removed from the partial lists.
+ *
+ * The slabs with the least items are placed last. This results in them
+ * being allocated from last increasing the chance that the last objects
+ * are freed in them.
+ */
+int kmem_cache_shrink(struct kmem_cache *s)
+{
+    int node;
+    int i;
+    struct kmem_cache_node *n;
+    struct page *page;
+    struct page *t;
+    int objects = oo_objects(s->max);
+    struct list_head *slabs_by_inuse =
+        kmalloc(sizeof(struct list_head) * objects, GFP_KERNEL);
+    unsigned long flags;
+
+    if (!slabs_by_inuse)
+        return -ENOMEM;
+
+    flush_all(s);
+    for_each_node_state(node, N_NORMAL_MEMORY) {
+        n = get_node(s, node);
+
+        if (!n->nr_partial)
+            continue;
+
+        for (i = 0; i < objects; i++)
+            INIT_LIST_HEAD(slabs_by_inuse + i);
+
+        spin_lock_irqsave(&n->list_lock, flags);
+
+        /*
+         * Build lists indexed by the items in use in each slab.
+         *
+         * Note that concurrent frees may occur while we hold the
+         * list_lock. page->inuse here is the upper limit.
+         */
+        list_for_each_entry_safe(page, t, &n->partial, lru) {
+            list_move(&page->lru, slabs_by_inuse + page->inuse);
+            if (!page->inuse)
+                n->nr_partial--;
+        }
+
+        /*
+         * Rebuild the partial list with the slabs filled up most
+         * first and the least used slabs at the end.
+         */
+        for (i = objects - 1; i > 0; i--)
+            list_splice(slabs_by_inuse + i, n->partial.prev);
+
+        spin_unlock_irqrestore(&n->list_lock, flags);
+
+        /* Release empty slabs */
+        list_for_each_entry_safe(page, t, slabs_by_inuse, lru)
+            discard_slab(s, page);
+    }
+
+    kfree(slabs_by_inuse);
+    return 0;
+}
+EXPORT_SYMBOL(kmem_cache_shrink);
+
+static int slab_mem_going_offline_callback(void *arg)
+{
+    struct kmem_cache *s;
+
+    mutex_lock(&slab_mutex);
+    list_for_each_entry(s, &slab_caches, list)
+        kmem_cache_shrink(s);
+    mutex_unlock(&slab_mutex);
+
+    return 0;
+}
+
+static void slab_mem_offline_callback(void *arg)
+{
+    struct kmem_cache_node *n;
+    struct kmem_cache *s;
+    struct memory_notify *marg = arg;
+    int offline_node;
+
+    offline_node = marg->status_change_nid_normal;
+
+    /*
+     * If the node still has available memory. we need kmem_cache_node
+     * for it yet.
+     */
+    if (offline_node < 0)
+        return;
+
+    mutex_lock(&slab_mutex);
+    list_for_each_entry(s, &slab_caches, list) {
+        n = get_node(s, offline_node);
+        if (n) {
+            /*
+             * if n->nr_slabs > 0, slabs still exist on the node
+             * that is going down. We were unable to free them,
+             * and offline_pages() function shouldn't call this
+             * callback. So, we must fail.
+             */
+            BUG_ON(slabs_node(s, offline_node));
+
+            s->node[offline_node] = NULL;
+            kmem_cache_free(kmem_cache_node, n);
+        }
+    }
+    mutex_unlock(&slab_mutex);
+}
+
+static int slab_mem_going_online_callback(void *arg)
+{
+    struct kmem_cache_node *n;
+    struct kmem_cache *s;
+    struct memory_notify *marg = arg;
+    int nid = marg->status_change_nid_normal;
+    int ret = 0;
+
+    /*
+     * If the node's memory is already available, then kmem_cache_node is
+     * already created. Nothing to do.
+     */
+    if (nid < 0)
+        return 0;
+
+    /*
+     * We are bringing a node online. No memory is available yet. We must
+     * allocate a kmem_cache_node structure in order to bring the node
+     * online.
+     */
+    mutex_lock(&slab_mutex);
+    list_for_each_entry(s, &slab_caches, list) {
+        /*
+         * XXX: kmem_cache_alloc_node will fallback to other nodes
+         *      since memory is not yet available from the node that
+         *      is brought up.
+         */
+        n = kmem_cache_alloc(kmem_cache_node, GFP_KERNEL);
+        if (!n) {
+            ret = -ENOMEM;
+            goto out;
+        }
+        init_kmem_cache_node(n);
+        s->node[nid] = n;
+    }
+out:
+    mutex_unlock(&slab_mutex);
+    return ret;
+}
+
+static int slab_memory_callback(struct notifier_block *self,
+                unsigned long action, void *arg)
+{
+    int ret = 0;
+
+    switch (action) {
+    case MEM_GOING_ONLINE:
+        ret = slab_mem_going_online_callback(arg);
+        break;
+    case MEM_GOING_OFFLINE:
+        ret = slab_mem_going_offline_callback(arg);
+        break;
+    case MEM_OFFLINE:
+    case MEM_CANCEL_ONLINE:
+        slab_mem_offline_callback(arg);
+        break;
+    case MEM_ONLINE:
+    case MEM_CANCEL_OFFLINE:
+        break;
+    }
+    if (ret)
+        ret = notifier_from_errno(ret);
+    else
+        ret = NOTIFY_OK;
+    return ret;
+}
+
+static struct notifier_block slab_memory_callback_nb = {
+    .notifier_call = slab_memory_callback,
+    .priority = SLAB_CALLBACK_PRI,
+};
+
+/********************************************************************
+ *            Basic setup of slabs
+ *******************************************************************/
+
+/*
+ * Used for early kmem_cache structures that were allocated using
+ * the page allocator. Allocate them properly then fix up the pointers
+ * that may be pointing to the wrong kmem_cache structure.
+ */
+
+static struct kmem_cache * __init bootstrap(struct kmem_cache *static_cache)
+{
+    int node;
+    struct kmem_cache *s = kmem_cache_zalloc(kmem_cache, GFP_NOWAIT);
+
+    memcpy(s, static_cache, kmem_cache->object_size);
+
+    /*
+     * This runs very early, and only the boot processor is supposed to be
+     * up.  Even if it weren't true, IRQs are not up so we couldn't fire
+     * IPIs around.
+     */
+    __flush_cpu_slab(s, smp_processor_id());
+    for_each_node_state(node, N_NORMAL_MEMORY) {
+        struct kmem_cache_node *n = get_node(s, node);
+        struct page *p;
+
+        if (n) {
+            list_for_each_entry(p, &n->partial, lru)
+                p->slab_cache = s;
+
+#ifdef CONFIG_SLEB_DEBUG
+            list_for_each_entry(p, &n->full, lru)
+                p->slab_cache = s;
+#endif
+        }
+    }
+    list_add(&s->list, &slab_caches);
+    return s;
+}
+
+void __init kmem_cache_init(void)
+{
+    static __initdata struct kmem_cache boot_kmem_cache,
+        boot_kmem_cache_node;
+
+    if (debug_guardpage_minorder())
+        slub_max_order = 0;
+
+    kmem_cache_node = &boot_kmem_cache_node;
+    kmem_cache = &boot_kmem_cache;
+
+    create_boot_cache(kmem_cache_node, "kmem_cache_node",
+        sizeof(struct kmem_cache_node), SLAB_HWCACHE_ALIGN);
+
+    register_hotmemory_notifier(&slab_memory_callback_nb);
+
+    /* Able to allocate the per node structures */
+    slab_state = PARTIAL;
+
+    create_boot_cache(kmem_cache, "kmem_cache",
+            offsetof(struct kmem_cache, node) +
+                nr_node_ids * sizeof(struct kmem_cache_node *),
+               SLAB_HWCACHE_ALIGN);
+
+    kmem_cache = bootstrap(&boot_kmem_cache);
+
+    /*
+     * Allocate kmem_cache_node properly from the kmem_cache slab.
+     * kmem_cache_node is separately allocated so no need to
+     * update any list pointers.
+     */
+    kmem_cache_node = bootstrap(&boot_kmem_cache_node);
+
+    /* Now we can use the kmem_cache to allocate kmalloc slabs */
+    create_kmalloc_caches(0);
+
+#ifdef CONFIG_SMP
+    register_cpu_notifier(&slab_notifier);
+#endif
+
+    printk(KERN_INFO
+        "SLUB: HWalign=%d, Order=%d-%d, MinObjects=%d,"
+        " CPUs=%d, Nodes=%d\n",
+        cache_line_size(),
+        slub_min_order, slub_max_order, slub_min_objects,
+        nr_cpu_ids, nr_node_ids);
+}
+
+void __init kmem_cache_init_late(void)
+{
+}
+
+/*
+ * Find a mergeable slab cache
+ */
+static int slab_unmergeable(struct kmem_cache *s)
+{
+    if (slub_nomerge || (s->flags & SLUB_NEVER_MERGE))
+        return 1;
+
+    if (s->ctor)
+        return 1;
+
+    /*
+     * We may have set a slab to be unmergeable during bootstrap.
+     */
+    if (s->refcount < 0)
+        return 1;
+
+    return 0;
+}
+
+static struct kmem_cache *find_mergeable(struct mem_cgroup *memcg, size_t size,
+        size_t align, unsigned long flags, const char *name,
+        void (*ctor)(void *))
+{
+    struct kmem_cache *s;
+
+    if (slub_nomerge || (flags & SLUB_NEVER_MERGE))
+        return NULL;
+
+    if (ctor)
+        return NULL;
+
+    size = ALIGN(size, sizeof(void *));
+    align = calculate_alignment(flags, align, size);
+    size = ALIGN(size, align);
+    flags = kmem_cache_flags(size, flags, name, NULL);
+
+    list_for_each_entry(s, &slab_caches, list) {
+        if (slab_unmergeable(s))
+            continue;
+
+        if (size > s->size)
+            continue;
+
+        if ((flags & SLUB_MERGE_SAME) != (s->flags & SLUB_MERGE_SAME))
+                continue;
+        /*
+         * Check if alignment is compatible.
+         * Courtesy of Adrian Drzewiecki
+         */
+        if ((s->size & ~(align - 1)) != s->size)
+            continue;
+
+        if (s->size - size >= sizeof(void *))
+            continue;
+
+        if (!cache_match_memcg(s, memcg))
+            continue;
+
+        return s;
+    }
+    return NULL;
+}
+
+struct kmem_cache *
+__kmem_cache_alias(struct mem_cgroup *memcg, const char *name, size_t size,
+           size_t align, unsigned long flags, void (*ctor)(void *))
+{
+    struct kmem_cache *s;
+
+    s = find_mergeable(memcg, size, align, flags, name, ctor);
+    if (s) {
+        s->refcount++;
+        /*
+         * Adjust the object sizes so that we clear
+         * the complete object on kzalloc.
+         */
+        s->object_size = max(s->object_size, (int)size);
+        s->inuse = max_t(int, s->inuse, ALIGN(size, sizeof(void *)));
+
+        if (sysfs_slab_alias(s, name)) {
+            s->refcount--;
+            s = NULL;
+        }
+    }
+
+    return s;
+}
+
+int __kmem_cache_create(struct kmem_cache *s, unsigned long flags)
+{
+    int err;
+
+    err = kmem_cache_open(s, flags);
+    if (err)
+        return err;
+
+    /* Mutex is not taken during early boot */
+    if (slab_state <= UP)
+        return 0;
+
+    memcg_propagate_slab_attrs(s);
+    mutex_unlock(&slab_mutex);
+    err = sysfs_slab_add(s);
+    mutex_lock(&slab_mutex);
+
+    if (err)
+        kmem_cache_close(s);
+
+    return err;
+}
+
+#ifdef CONFIG_SMP
+/*
+ * Use the cpu notifier to insure that the cpu slabs are flushed when
+ * necessary.
+ */
+static int __cpuinit slab_cpuup_callback(struct notifier_block *nfb,
+        unsigned long action, void *hcpu)
+{
+    long cpu = (long)hcpu;
+    struct kmem_cache *s;
+    unsigned long flags;
+
+    switch (action) {
+    case CPU_UP_CANCELED:
+    case CPU_UP_CANCELED_FROZEN:
+    case CPU_DEAD:
+    case CPU_DEAD_FROZEN:
+        mutex_lock(&slab_mutex);
+        list_for_each_entry(s, &slab_caches, list) {
+            local_irq_save(flags);
+            __flush_cpu_slab(s, cpu);
+            local_irq_restore(flags);
+        }
+        mutex_unlock(&slab_mutex);
+        break;
+    default:
+        break;
+    }
+    return NOTIFY_OK;
+}
+
+static struct notifier_block __cpuinitdata slab_notifier = {
+    .notifier_call = slab_cpuup_callback
+};
+
+#endif
+
+void *__kmalloc_track_caller(size_t size, gfp_t gfpflags, unsigned long caller)
+{
+    struct kmem_cache *s;
+    void *ret;
+
+    if (unlikely(size > KMALLOC_MAX_CACHE_SIZE))
+        return kmalloc_large(size, gfpflags);
+
+    s = kmalloc_slab(size, gfpflags);
+
+    if (unlikely(ZERO_OR_NULL_PTR(s)))
+        return s;
+
+    ret = slab_alloc(s, gfpflags, caller);
+
+    /* Honor the call site pointer we received. */
+    trace_kmalloc(caller, ret, size, s->size, gfpflags);
+
+    return ret;
+}
+
+#ifdef CONFIG_NUMA
+void *__kmalloc_node_track_caller(size_t size, gfp_t gfpflags,
+                    int node, unsigned long caller)
+{
+    struct kmem_cache *s;
+    void *ret;
+
+    if (unlikely(size > KMALLOC_MAX_CACHE_SIZE)) {
+        ret = kmalloc_large_node(size, gfpflags, node);
+
+        trace_kmalloc_node(caller, ret,
+                   size, PAGE_SIZE << get_order(size),
+                   gfpflags, node);
+
+        return ret;
+    }
+
+    s = kmalloc_slab(size, gfpflags);
+
+    if (unlikely(ZERO_OR_NULL_PTR(s)))
+        return s;
+
+    ret = slab_alloc_node(s, gfpflags, node, caller);
+
+    /* Honor the call site pointer we received. */
+    trace_kmalloc_node(caller, ret, size, s->size, gfpflags, node);
+
+    return ret;
+}
+#endif
+
+#ifdef CONFIG_SYSFS
+static int count_inuse(struct page *page)
+{
+    return page->inuse;
+}
+
+static int count_total(struct page *page)
+{
+    return page->objects;
+}
+#endif
+
+#ifdef CONFIG_SLEB_DEBUG
+static int validate_slab(struct kmem_cache *s, struct page *page,
+                        unsigned long *map)
+{
+    void *p;
+    void *addr = page_address(page);
+
+    if (!check_slab(s, page) ||
+            !on_freelist(s, page, NULL))
+        return 0;
+
+    /* Now we know that a valid freelist exists */
+    bitmap_zero(map, page->objects);
+
+    get_map(s, page, map);
+    for_each_object(p, s, addr, page->objects) {
+        if (test_bit(slab_index(p, s, addr), map))
+            if (!check_object(s, page, p, SLUB_RED_INACTIVE))
+                return 0;
+    }
+
+    for_each_object(p, s, addr, page->objects)
+        if (!test_bit(slab_index(p, s, addr), map))
+            if (!check_object(s, page, p, SLUB_RED_ACTIVE))
+                return 0;
+    return 1;
+}
+
+static void validate_slab_slab(struct kmem_cache *s, struct page *page,
+                        unsigned long *map)
+{
+    slab_lock(page);
+    validate_slab(s, page, map);
+    slab_unlock(page);
+}
+
+static int validate_slab_node(struct kmem_cache *s,
+        struct kmem_cache_node *n, unsigned long *map)
+{
+    unsigned long count = 0;
+    struct page *page;
+    unsigned long flags;
+
+    spin_lock_irqsave(&n->list_lock, flags);
+
+    list_for_each_entry(page, &n->partial, lru) {
+        validate_slab_slab(s, page, map);
+        count++;
+    }
+    if (count != n->nr_partial)
+        printk(KERN_ERR "SLUB %s: %ld partial slabs counted but "
+            "counter=%ld\n", s->name, count, n->nr_partial);
+
+    if (!(s->flags & SLAB_STORE_USER))
+        goto out;
+
+    list_for_each_entry(page, &n->full, lru) {
+        validate_slab_slab(s, page, map);
+        count++;
+    }
+    if (count != atomic_long_read(&n->nr_slabs))
+        printk(KERN_ERR "SLUB: %s %ld slabs counted but "
+            "counter=%ld\n", s->name, count,
+            atomic_long_read(&n->nr_slabs));
+
+out:
+    spin_unlock_irqrestore(&n->list_lock, flags);
+    return count;
+}
+
+static long validate_slab_cache(struct kmem_cache *s)
+{
+    int node;
+    unsigned long count = 0;
+    unsigned long *map = kmalloc(BITS_TO_LONGS(oo_objects(s->max)) *
+                sizeof(unsigned long), GFP_KERNEL);
+
+    if (!map)
+        return -ENOMEM;
+
+    flush_all(s);
+    for_each_node_state(node, N_NORMAL_MEMORY) {
+        struct kmem_cache_node *n = get_node(s, node);
+
+        count += validate_slab_node(s, n, map);
+    }
+    kfree(map);
+    return count;
+}
+/*
+ * Generate lists of code addresses where slabcache objects are allocated
+ * and freed.
+ */
+
+struct location {
+    unsigned long count;
+    unsigned long addr;
+    long long sum_time;
+    long min_time;
+    long max_time;
+    long min_pid;
+    long max_pid;
+    DECLARE_BITMAP(cpus, NR_CPUS);
+    nodemask_t nodes;
+};
+
+struct loc_track {
+    unsigned long max;
+    unsigned long count;
+    struct location *loc;
+};
+
+static void free_loc_track(struct loc_track *t)
+{
+    if (t->max)
+        free_pages((unsigned long)t->loc,
+            get_order(sizeof(struct location) * t->max));
+}
+
+static int alloc_loc_track(struct loc_track *t, unsigned long max, gfp_t flags)
+{
+    struct location *l;
+    int order;
+
+    order = get_order(sizeof(struct location) * max);
+
+    l = (void *)__get_free_pages(flags, order);
+    if (!l)
+        return 0;
+
+    if (t->count) {
+        memcpy(l, t->loc, sizeof(struct location) * t->count);
+        free_loc_track(t);
+    }
+    t->max = max;
+    t->loc = l;
+    return 1;
+}
+
+static int add_location(struct loc_track *t, struct kmem_cache *s,
+                const struct track *track)
+{
+    long start, end, pos;
+    struct location *l;
+    unsigned long caddr;
+    unsigned long age = jiffies - track->when;
+
+    start = -1;
+    end = t->count;
+
+    for ( ; ; ) {
+        pos = start + (end - start + 1) / 2;
+
+        /*
+         * There is nothing at "end". If we end up there
+         * we need to add something to before end.
+         */
+        if (pos == end)
+            break;
+
+        caddr = t->loc[pos].addr;
+        if (track->addr == caddr) {
+
+            l = &t->loc[pos];
+            l->count++;
+            if (track->when) {
+                l->sum_time += age;
+                if (age < l->min_time)
+                    l->min_time = age;
+                if (age > l->max_time)
+                    l->max_time = age;
+
+                if (track->pid < l->min_pid)
+                    l->min_pid = track->pid;
+                if (track->pid > l->max_pid)
+                    l->max_pid = track->pid;
+
+                cpumask_set_cpu(track->cpu,
+                        to_cpumask(l->cpus));
+            }
+            node_set(page_to_nid(virt_to_page(track)), l->nodes);
+            return 1;
+        }
+
+        if (track->addr < caddr)
+            end = pos;
+        else
+            start = pos;
+    }
+
+    /*
+     * Not found. Insert new tracking element.
+     */
+    if (t->count >= t->max && !alloc_loc_track(t, 2 * t->max, GFP_ATOMIC))
+        return 0;
+
+    l = t->loc + pos;
+    if (pos < t->count)
+        memmove(l + 1, l,
+            (t->count - pos) * sizeof(struct location));
+    t->count++;
+    l->count = 1;
+    l->addr = track->addr;
+    l->sum_time = age;
+    l->min_time = age;
+    l->max_time = age;
+    l->min_pid = track->pid;
+    l->max_pid = track->pid;
+    cpumask_clear(to_cpumask(l->cpus));
+    cpumask_set_cpu(track->cpu, to_cpumask(l->cpus));
+    nodes_clear(l->nodes);
+    node_set(page_to_nid(virt_to_page(track)), l->nodes);
+    return 1;
+}
+
+static void process_slab(struct loc_track *t, struct kmem_cache *s,
+        struct page *page, enum track_item alloc,
+        unsigned long *map)
+{
+    void *addr = page_address(page);
+    void *p;
+
+    bitmap_zero(map, page->objects);
+    get_map(s, page, map);
+
+    for_each_object(p, s, addr, page->objects)
+        if (!test_bit(slab_index(p, s, addr), map))
+            add_location(t, s, get_track(s, p, alloc));
+}
+
+static int list_locations(struct kmem_cache *s, char *buf,
+                    enum track_item alloc)
+{
+    int len = 0;
+    unsigned long i;
+    struct loc_track t = { 0, 0, NULL };
+    int node;
+    unsigned long *map = kmalloc(BITS_TO_LONGS(oo_objects(s->max)) *
+                     sizeof(unsigned long), GFP_KERNEL);
+
+    if (!map || !alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location),
+                     GFP_TEMPORARY)) {
+        kfree(map);
+        return sprintf(buf, "Out of memory\n");
+    }
+    /* Push back cpu slabs */
+    flush_all(s);
+
+    for_each_node_state(node, N_NORMAL_MEMORY) {
+        struct kmem_cache_node *n = get_node(s, node);
+        unsigned long flags;
+        struct page *page;
+
+        if (!atomic_long_read(&n->nr_slabs))
+            continue;
+
+        spin_lock_irqsave(&n->list_lock, flags);
+        list_for_each_entry(page, &n->partial, lru)
+            process_slab(&t, s, page, alloc, map);
+        list_for_each_entry(page, &n->full, lru)
+            process_slab(&t, s, page, alloc, map);
+        spin_unlock_irqrestore(&n->list_lock, flags);
+    }
+
+    for (i = 0; i < t.count; i++) {
+        struct location *l = &t.loc[i];
+
+        if (len > PAGE_SIZE - KSYM_SYMBOL_LEN - 100)
+            break;
+        len += sprintf(buf + len, "%7ld ", l->count);
+
+        if (l->addr)
+            len += sprintf(buf + len, "%pS", (void *)l->addr);
+        else
+            len += sprintf(buf + len, "<not-available>");
+
+        if (l->sum_time != l->min_time) {
+            len += sprintf(buf + len, " age=%ld/%ld/%ld",
+                l->min_time,
+                (long)div_u64(l->sum_time, l->count),
+                l->max_time);
+        } else
+            len += sprintf(buf + len, " age=%ld",
+                l->min_time);
+
+        if (l->min_pid != l->max_pid)
+            len += sprintf(buf + len, " pid=%ld-%ld",
+                l->min_pid, l->max_pid);
+        else
+            len += sprintf(buf + len, " pid=%ld",
+                l->min_pid);
+
+        if (num_online_cpus() > 1 &&
+                !cpumask_empty(to_cpumask(l->cpus)) &&
+                len < PAGE_SIZE - 60) {
+            len += sprintf(buf + len, " cpus=");
+            len += cpulist_scnprintf(buf + len, PAGE_SIZE - len - 50,
+                         to_cpumask(l->cpus));
+        }
+
+        if (nr_online_nodes > 1 && !nodes_empty(l->nodes) &&
+                len < PAGE_SIZE - 60) {
+            len += sprintf(buf + len, " nodes=");
+            len += nodelist_scnprintf(buf + len, PAGE_SIZE - len - 50,
+                    l->nodes);
+        }
+
+        len += sprintf(buf + len, "\n");
+    }
+
+    free_loc_track(&t);
+    kfree(map);
+    if (!t.count)
+        len += sprintf(buf, "No data\n");
+    return len;
+}
+#endif
+
+#ifdef SLUB_RESILIENCY_TEST
+static void resiliency_test(void)
+{
+    u8 *p;
+
+    BUILD_BUG_ON(KMALLOC_MIN_SIZE > 16 || KMALLOC_SHIFT_HIGH < 10);
+
+    printk(KERN_ERR "SLUB resiliency testing\n");
+    printk(KERN_ERR "-----------------------\n");
+    printk(KERN_ERR "A. Corruption after allocation\n");
+
+    p = kzalloc(16, GFP_KERNEL);
+    p[16] = 0x12;
+    printk(KERN_ERR "\n1. kmalloc-16: Clobber Redzone/next pointer"
+            " 0x12->0x%p\n\n", p + 16);
+
+    validate_slab_cache(kmalloc_caches[4]);
+
+    /* Hmmm... The next two are dangerous */
+    p = kzalloc(32, GFP_KERNEL);
+    p[32 + sizeof(void *)] = 0x34;
+    printk(KERN_ERR "\n2. kmalloc-32: Clobber next pointer/next slab"
+            " 0x34 -> -0x%p\n", p);
+    printk(KERN_ERR
+        "If allocated object is overwritten then not detectable\n\n");
+
+    validate_slab_cache(kmalloc_caches[5]);
+    p = kzalloc(64, GFP_KERNEL);
+    p += 64 + (get_cycles() & 0xff) * sizeof(void *);
+    *p = 0x56;
+    printk(KERN_ERR "\n3. kmalloc-64: corrupting random byte 0x56->0x%p\n",
+                                    p);
+    printk(KERN_ERR
+        "If allocated object is overwritten then not detectable\n\n");
+    validate_slab_cache(kmalloc_caches[6]);
+
+    printk(KERN_ERR "\nB. Corruption after free\n");
+    p = kzalloc(128, GFP_KERNEL);
+    kfree(p);
+    *p = 0x78;
+    printk(KERN_ERR "1. kmalloc-128: Clobber first word 0x78->0x%p\n\n", p);
+    validate_slab_cache(kmalloc_caches[7]);
+
+    p = kzalloc(256, GFP_KERNEL);
+    kfree(p);
+    p[50] = 0x9a;
+    printk(KERN_ERR "\n2. kmalloc-256: Clobber 50th byte 0x9a->0x%p\n\n",
+            p);
+    validate_slab_cache(kmalloc_caches[8]);
+
+    p = kzalloc(512, GFP_KERNEL);
+    kfree(p);
+    p[512] = 0xab;
+    printk(KERN_ERR "\n3. kmalloc-512: Clobber redzone 0xab->0x%p\n\n", p);
+    validate_slab_cache(kmalloc_caches[9]);
+}
+#else
+#ifdef CONFIG_SYSFS
+static void resiliency_test(void) {};
+#endif
+#endif
+
+#ifdef CONFIG_SYSFS
+enum slab_stat_type {
+    SL_ALL,            /* All slabs */
+    SL_PARTIAL,        /* Only partially allocated slabs */
+    SL_CPU,            /* Only slabs used for cpu caches */
+    SL_OBJECTS,        /* Determine allocated objects not slabs */
+    SL_TOTAL        /* Determine object capacity not slabs */
+};
+
+#define SO_ALL        (1 << SL_ALL)
+#define SO_PARTIAL    (1 << SL_PARTIAL)
+#define SO_CPU        (1 << SL_CPU)
+#define SO_OBJECTS    (1 << SL_OBJECTS)
+#define SO_TOTAL    (1 << SL_TOTAL)
+
+static ssize_t show_slab_objects(struct kmem_cache *s,
+                char *buf, unsigned long flags)
+{
+    unsigned long total = 0;
+    int node;
+    int x;
+    unsigned long *nodes;
+    unsigned long *per_cpu;
+
+    nodes = kzalloc(2 * sizeof(unsigned long) * nr_node_ids, GFP_KERNEL);
+    if (!nodes)
+        return -ENOMEM;
+    per_cpu = nodes + nr_node_ids;
+
+    if (flags & SO_CPU) {
+        int cpu;
+
+        for_each_possible_cpu(cpu) {
+            struct kmem_cache_cpu *c = per_cpu_ptr(s->cpu_slab, cpu);
+            int node;
+            struct page *page;
+
+            page = ACCESS_ONCE(c->page);
+            if (!page)
+                continue;
+
+            node = page_to_nid(page);
+            if (flags & SO_TOTAL)
+                x = page->objects;
+            else if (flags & SO_OBJECTS)
+                x = page->inuse;
+            else
+                x = 1;
+
+            total += x;
+            nodes[node] += x;
+
+            page = ACCESS_ONCE(c->partial);
+            if (page) {
+                x = page->pobjects;
+                total += x;
+                nodes[node] += x;
+            }
+
+            per_cpu[node]++;
+        }
+    }
+
+    lock_memory_hotplug();
+#ifdef CONFIG_SLEB_DEBUG
+    if (flags & SO_ALL) {
+        for_each_node_state(node, N_NORMAL_MEMORY) {
+            struct kmem_cache_node *n = get_node(s, node);
+
+        if (flags & SO_TOTAL)
+            x = atomic_long_read(&n->total_objects);
+        else if (flags & SO_OBJECTS)
+            x = atomic_long_read(&n->total_objects) -
+                count_partial(n, count_free);
+
+            else
+                x = atomic_long_read(&n->nr_slabs);
+            total += x;
+            nodes[node] += x;
+        }
+
+    } else
+#endif
+    if (flags & SO_PARTIAL) {
+        for_each_node_state(node, N_NORMAL_MEMORY) {
+            struct kmem_cache_node *n = get_node(s, node);
+
+            if (flags & SO_TOTAL)
+                x = count_partial(n, count_total);
+            else if (flags & SO_OBJECTS)
+                x = count_partial(n, count_inuse);
+            else
+                x = n->nr_partial;
+            total += x;
+            nodes[node] += x;
+        }
+    }
+    x = sprintf(buf, "%lu", total);
+#ifdef CONFIG_NUMA
+    for_each_node_state(node, N_NORMAL_MEMORY)
+        if (nodes[node])
+            x += sprintf(buf + x, " N%d=%lu",
+                    node, nodes[node]);
+#endif
+    unlock_memory_hotplug();
+    kfree(nodes);
+    return x + sprintf(buf + x, "\n");
+}
+
+#ifdef CONFIG_SLEB_DEBUG
+static int any_slab_objects(struct kmem_cache *s)
+{
+    int node;
+
+    for_each_online_node(node) {
+        struct kmem_cache_node *n = get_node(s, node);
+
+        if (!n)
+            continue;
+
+        if (atomic_long_read(&n->total_objects))
+            return 1;
+    }
+    return 0;
+}
+#endif
+
+#define to_slab_attr(n) container_of(n, struct slab_attribute, attr)
+#define to_slab(n) container_of(n, struct kmem_cache, kobj)
+
+struct slab_attribute {
+    struct attribute attr;
+    ssize_t (*show)(struct kmem_cache *s, char *buf);
+    ssize_t (*store)(struct kmem_cache *s, const char *x, size_t count);
+};
+
+#define SLAB_ATTR_RO(_name) \
+    static struct slab_attribute _name##_attr = \
+    __ATTR(_name, 0400, _name##_show, NULL)
+
+#define SLAB_ATTR(_name) \
+    static struct slab_attribute _name##_attr =  \
+    __ATTR(_name, 0600, _name##_show, _name##_store)
+
+static ssize_t slab_size_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", s->size);
+}
+SLAB_ATTR_RO(slab_size);
+
+static ssize_t align_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", s->align);
+}
+SLAB_ATTR_RO(align);
+
+static ssize_t object_size_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", s->object_size);
+}
+SLAB_ATTR_RO(object_size);
+
+static ssize_t objs_per_slab_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", oo_objects(s->oo));
+}
+SLAB_ATTR_RO(objs_per_slab);
+
+static ssize_t order_store(struct kmem_cache *s,
+                const char *buf, size_t length)
+{
+    unsigned long order;
+    int err;
+
+    err = strict_strtoul(buf, 10, &order);
+    if (err)
+        return err;
+
+    if (order > slub_max_order || order < slub_min_order)
+        return -EINVAL;
+
+    calculate_sizes(s, order);
+    return length;
+}
+
+static ssize_t order_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", oo_order(s->oo));
+}
+SLAB_ATTR(order);
+
+static ssize_t min_partial_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%lu\n", s->min_partial);
+}
+
+static ssize_t min_partial_store(struct kmem_cache *s, const char *buf,
+                 size_t length)
+{
+    unsigned long min;
+    int err;
+
+    err = strict_strtoul(buf, 10, &min);
+    if (err)
+        return err;
+
+    set_min_partial(s, min);
+    return length;
+}
+SLAB_ATTR(min_partial);
+
+static ssize_t cpu_partial_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%u\n", s->cpu_partial);
+}
+
+static ssize_t cpu_partial_store(struct kmem_cache *s, const char *buf,
+                 size_t length)
+{
+    unsigned long objects;
+    int err;
+
+    err = strict_strtoul(buf, 10, &objects);
+    if (err)
+        return err;
+    if (objects && kmem_cache_debug(s))
+        return -EINVAL;
+
+    s->cpu_partial = objects;
+    flush_all(s);
+    return length;
+}
+SLAB_ATTR(cpu_partial);
+
+static ssize_t ctor_show(struct kmem_cache *s, char *buf)
+{
+    if (!s->ctor)
+        return 0;
+    return sprintf(buf, "%pS\n", s->ctor);
+}
+SLAB_ATTR_RO(ctor);
+
+static ssize_t aliases_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", s->refcount - 1);
+}
+SLAB_ATTR_RO(aliases);
+
+static ssize_t partial_show(struct kmem_cache *s, char *buf)
+{
+    return show_slab_objects(s, buf, SO_PARTIAL);
+}
+SLAB_ATTR_RO(partial);
+
+static ssize_t cpu_slabs_show(struct kmem_cache *s, char *buf)
+{
+    return show_slab_objects(s, buf, SO_CPU);
+}
+SLAB_ATTR_RO(cpu_slabs);
+
+static ssize_t objects_show(struct kmem_cache *s, char *buf)
+{
+    return show_slab_objects(s, buf, SO_ALL|SO_OBJECTS);
+}
+SLAB_ATTR_RO(objects);
+
+static ssize_t objects_partial_show(struct kmem_cache *s, char *buf)
+{
+    return show_slab_objects(s, buf, SO_PARTIAL|SO_OBJECTS);
+}
+SLAB_ATTR_RO(objects_partial);
+
+static ssize_t slabs_cpu_partial_show(struct kmem_cache *s, char *buf)
+{
+    int objects = 0;
+    int pages = 0;
+    int cpu;
+    int len;
+
+    for_each_online_cpu(cpu) {
+        struct page *page = per_cpu_ptr(s->cpu_slab, cpu)->partial;
+
+        if (page) {
+            pages += page->pages;
+            objects += page->pobjects;
+        }
+    }
+
+    len = sprintf(buf, "%d(%d)", objects, pages);
+
+#ifdef CONFIG_SMP
+    for_each_online_cpu(cpu) {
+        struct page *page = per_cpu_ptr(s->cpu_slab, cpu) ->partial;
+
+        if (page && len < PAGE_SIZE - 20)
+            len += sprintf(buf + len, " C%d=%d(%d)", cpu,
+                page->pobjects, page->pages);
+    }
+#endif
+    return len + sprintf(buf + len, "\n");
+}
+SLAB_ATTR_RO(slabs_cpu_partial);
+
+static ssize_t reclaim_account_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_RECLAIM_ACCOUNT));
+}
+
+static ssize_t reclaim_account_store(struct kmem_cache *s,
+                const char *buf, size_t length)
+{
+    s->flags &= ~SLAB_RECLAIM_ACCOUNT;
+    if (buf[0] == '1')
+        s->flags |= SLAB_RECLAIM_ACCOUNT;
+    return length;
+}
+SLAB_ATTR(reclaim_account);
+
+static ssize_t hwcache_align_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_HWCACHE_ALIGN));
+}
+SLAB_ATTR_RO(hwcache_align);
+
+#ifdef CONFIG_ZONE_DMA
+static ssize_t cache_dma_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_CACHE_DMA));
+}
+SLAB_ATTR_RO(cache_dma);
+#endif
+
+static ssize_t destroy_by_rcu_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_DESTROY_BY_RCU));
+}
+SLAB_ATTR_RO(destroy_by_rcu);
+
+static ssize_t reserved_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", s->reserved);
+}
+SLAB_ATTR_RO(reserved);
+
+#ifdef CONFIG_SLEB_DEBUG
+static ssize_t slabs_show(struct kmem_cache *s, char *buf)
+{
+    return show_slab_objects(s, buf, SO_ALL);
+}
+SLAB_ATTR_RO(slabs);
+
+static ssize_t total_objects_show(struct kmem_cache *s, char *buf)
+{
+    return show_slab_objects(s, buf, SO_ALL|SO_TOTAL);
+}
+SLAB_ATTR_RO(total_objects);
+
+static ssize_t sanity_checks_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_DEBUG_FREE));
+}
+
+static ssize_t sanity_checks_store(struct kmem_cache *s,
+                const char *buf, size_t length)
+{
+    s->flags &= ~SLAB_DEBUG_FREE;
+    if (buf[0] == '1') {
+        s->flags &= ~__CMPXCHG_DOUBLE;
+        s->flags |= SLAB_DEBUG_FREE;
+    }
+    return length;
+}
+SLAB_ATTR(sanity_checks);
+
+static ssize_t trace_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_TRACE));
+}
+
+static ssize_t trace_store(struct kmem_cache *s, const char *buf,
+                            size_t length)
+{
+    s->flags &= ~SLAB_TRACE;
+    if (buf[0] == '1') {
+        s->flags &= ~__CMPXCHG_DOUBLE;
+        s->flags |= SLAB_TRACE;
+    }
+    return length;
+}
+SLAB_ATTR(trace);
+
+static ssize_t red_zone_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_RED_ZONE));
+}
+
+static ssize_t red_zone_store(struct kmem_cache *s,
+                const char *buf, size_t length)
+{
+    if (any_slab_objects(s))
+        return -EBUSY;
+
+    s->flags &= ~SLAB_RED_ZONE;
+    if (buf[0] == '1') {
+        s->flags &= ~__CMPXCHG_DOUBLE;
+        s->flags |= SLAB_RED_ZONE;
+    }
+    calculate_sizes(s, -1);
+    return length;
+}
+SLAB_ATTR(red_zone);
+
+static ssize_t poison_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_POISON));
+}
+
+static ssize_t poison_store(struct kmem_cache *s,
+                const char *buf, size_t length)
+{
+    if (any_slab_objects(s))
+        return -EBUSY;
+
+    s->flags &= ~SLAB_POISON;
+    if (buf[0] == '1') {
+        s->flags &= ~__CMPXCHG_DOUBLE;
+        s->flags |= SLAB_POISON;
+    }
+    calculate_sizes(s, -1);
+    return length;
+}
+SLAB_ATTR(poison);
+
+static ssize_t store_user_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", !!(s->flags & SLAB_STORE_USER));
+}
+
+static ssize_t store_user_store(struct kmem_cache *s,
+                const char *buf, size_t length)
+{
+    if (any_slab_objects(s))
+        return -EBUSY;
+
+    s->flags &= ~SLAB_STORE_USER;
+    if (buf[0] == '1') {
+        s->flags &= ~__CMPXCHG_DOUBLE;
+        s->flags |= SLAB_STORE_USER;
+    }
+    calculate_sizes(s, -1);
+    return length;
+}
+SLAB_ATTR(store_user);
+
+static ssize_t validate_show(struct kmem_cache *s, char *buf)
+{
+    return 0;
+}
+
+static ssize_t validate_store(struct kmem_cache *s,
+            const char *buf, size_t length)
+{
+    int ret = -EINVAL;
+
+    if (buf[0] == '1') {
+        ret = validate_slab_cache(s);
+        if (ret >= 0)
+            ret = length;
+    }
+    return ret;
+}
+SLAB_ATTR(validate);
+
+static ssize_t alloc_calls_show(struct kmem_cache *s, char *buf)
+{
+    if (!(s->flags & SLAB_STORE_USER))
+        return -ENOSYS;
+    return list_locations(s, buf, TRACK_ALLOC);
+}
+SLAB_ATTR_RO(alloc_calls);
+
+static ssize_t free_calls_show(struct kmem_cache *s, char *buf)
+{
+    if (!(s->flags & SLAB_STORE_USER))
+        return -ENOSYS;
+    return list_locations(s, buf, TRACK_FREE);
+}
+SLAB_ATTR_RO(free_calls);
+#endif /* CONFIG_SLEB_DEBUG */
+
+#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 shrink_show(struct kmem_cache *s, char *buf)
+{
+    return 0;
+}
+
+static ssize_t shrink_store(struct kmem_cache *s,
+            const char *buf, size_t length)
+{
+    if (buf[0] == '1') {
+        int rc = kmem_cache_shrink(s);
+
+        if (rc)
+            return rc;
+    } else
+        return -EINVAL;
+    return length;
+}
+SLAB_ATTR(shrink);
+
+#ifdef CONFIG_NUMA
+static ssize_t remote_node_defrag_ratio_show(struct kmem_cache *s, char *buf)
+{
+    return sprintf(buf, "%d\n", s->remote_node_defrag_ratio / 10);
+}
+
+static ssize_t remote_node_defrag_ratio_store(struct kmem_cache *s,
+                const char *buf, size_t length)
+{
+    unsigned long ratio;
+    int err;
+
+    err = strict_strtoul(buf, 10, &ratio);
+    if (err)
+        return err;
+
+    if (ratio <= 100)
+        s->remote_node_defrag_ratio = ratio * 10;
+
+    return length;
+}
+SLAB_ATTR(remote_node_defrag_ratio);
+#endif
+
+#ifdef CONFIG_SLEB_STATS
+static int show_stat(struct kmem_cache *s, char *buf, enum stat_item si)
+{
+    unsigned long sum  = 0;
+    int cpu;
+    int len;
+    int *data = kmalloc(nr_cpu_ids * sizeof(int), GFP_KERNEL);
+
+    if (!data)
+        return -ENOMEM;
+
+    for_each_online_cpu(cpu) {
+        unsigned x = per_cpu_ptr(s->cpu_slab, cpu)->stat[si];
+
+        data[cpu] = x;
+        sum += x;
+    }
+
+    len = sprintf(buf, "%lu", sum);
+
+#ifdef CONFIG_SMP
+    for_each_online_cpu(cpu) {
+        if (data[cpu] && len < PAGE_SIZE - 20)
+            len += sprintf(buf + len, " C%d=%u", cpu, data[cpu]);
+    }
+#endif
+    kfree(data);
+    return len + sprintf(buf + len, "\n");
+}
+
+static void clear_stat(struct kmem_cache *s, enum stat_item si)
+{
+    int cpu;
+
+    for_each_online_cpu(cpu)
+        per_cpu_ptr(s->cpu_slab, cpu)->stat[si] = 0;
+}
+
+#define STAT_ATTR(si, text)                     \
+static ssize_t text##_show(struct kmem_cache *s, char *buf)    \
+{                                \
+    return show_stat(s, buf, si);                \
+}                                \
+static ssize_t text##_store(struct kmem_cache *s,        \
+                const char *buf, size_t length)    \
+{                                \
+    if (buf[0] != '0')                    \
+        return -EINVAL;                    \
+    clear_stat(s, si);                    \
+    return length;                        \
+}                                \
+SLAB_ATTR(text);                        \
+
+STAT_ATTR(ALLOC_FASTPATH, alloc_fastpath);
+STAT_ATTR(ALLOC_SLOWPATH, alloc_slowpath);
+STAT_ATTR(FREE_FASTPATH, free_fastpath);
+STAT_ATTR(FREE_SLOWPATH, free_slowpath);
+STAT_ATTR(FREE_FROZEN, free_frozen);
+STAT_ATTR(FREE_ADD_PARTIAL, free_add_partial);
+STAT_ATTR(FREE_REMOVE_PARTIAL, free_remove_partial);
+STAT_ATTR(ALLOC_FROM_PARTIAL, alloc_from_partial);
+STAT_ATTR(ALLOC_SLAB, alloc_slab);
+STAT_ATTR(ALLOC_REFILL, alloc_refill);
+STAT_ATTR(ALLOC_NODE_MISMATCH, alloc_node_mismatch);
+STAT_ATTR(FREE_SLAB, free_slab);
+STAT_ATTR(CPUSLAB_FLUSH, cpuslab_flush);
+STAT_ATTR(DEACTIVATE_FULL, deactivate_full);
+STAT_ATTR(DEACTIVATE_EMPTY, deactivate_empty);
+STAT_ATTR(DEACTIVATE_TO_HEAD, deactivate_to_head);
+STAT_ATTR(DEACTIVATE_TO_TAIL, deactivate_to_tail);
+STAT_ATTR(DEACTIVATE_REMOTE_FREES, deactivate_remote_frees);
+STAT_ATTR(DEACTIVATE_BYPASS, deactivate_bypass);
+STAT_ATTR(ORDER_FALLBACK, order_fallback);
+STAT_ATTR(CMPXCHG_DOUBLE_CPU_FAIL, cmpxchg_double_cpu_fail);
+STAT_ATTR(CMPXCHG_DOUBLE_FAIL, cmpxchg_double_fail);
+STAT_ATTR(CPU_PARTIAL_ALLOC, cpu_partial_alloc);
+STAT_ATTR(CPU_PARTIAL_FREE, cpu_partial_free);
+STAT_ATTR(CPU_PARTIAL_NODE, cpu_partial_node);
+STAT_ATTR(CPU_PARTIAL_DRAIN, cpu_partial_drain);
+#endif
+
+static struct attribute *slab_attrs[] = {
+    &slab_size_attr.attr,
+    &object_size_attr.attr,
+    &objs_per_slab_attr.attr,
+    &order_attr.attr,
+    &min_partial_attr.attr,
+    &cpu_partial_attr.attr,
+    &objects_attr.attr,
+    &objects_partial_attr.attr,
+    &partial_attr.attr,
+    &cpu_slabs_attr.attr,
+    &ctor_attr.attr,
+    &aliases_attr.attr,
+    &align_attr.attr,
+    &hwcache_align_attr.attr,
+    &reclaim_account_attr.attr,
+    &destroy_by_rcu_attr.attr,
+    &shrink_attr.attr,
+    &reserved_attr.attr,
+    &slabs_cpu_partial_attr.attr,
+#ifdef CONFIG_SLEB_DEBUG
+    &total_objects_attr.attr,
+    &slabs_attr.attr,
+    &sanity_checks_attr.attr,
+    &trace_attr.attr,
+    &red_zone_attr.attr,
+    &poison_attr.attr,
+    &store_user_attr.attr,
+    &validate_attr.attr,
+    &alloc_calls_attr.attr,
+    &free_calls_attr.attr,
+#endif
+#ifdef CONFIG_ZONE_DMA
+    &cache_dma_attr.attr,
+#endif
+#ifdef CONFIG_NUMA
+    &remote_node_defrag_ratio_attr.attr,
+#endif
+#ifdef CONFIG_SLEB_STATS
+    &alloc_fastpath_attr.attr,
+    &alloc_slowpath_attr.attr,
+    &free_fastpath_attr.attr,
+    &free_slowpath_attr.attr,
+    &free_frozen_attr.attr,
+    &free_add_partial_attr.attr,
+    &free_remove_partial_attr.attr,
+    &alloc_from_partial_attr.attr,
+    &alloc_slab_attr.attr,
+    &alloc_refill_attr.attr,
+    &alloc_node_mismatch_attr.attr,
+    &free_slab_attr.attr,
+    &cpuslab_flush_attr.attr,
+    &deactivate_full_attr.attr,
+    &deactivate_empty_attr.attr,
+    &deactivate_to_head_attr.attr,
+    &deactivate_to_tail_attr.attr,
+    &deactivate_remote_frees_attr.attr,
+    &deactivate_bypass_attr.attr,
+    &order_fallback_attr.attr,
+    &cmpxchg_double_fail_attr.attr,
+    &cmpxchg_double_cpu_fail_attr.attr,
+    &cpu_partial_alloc_attr.attr,
+    &cpu_partial_free_attr.attr,
+    &cpu_partial_node_attr.attr,
+    &cpu_partial_drain_attr.attr,
+#endif
+#ifdef CONFIG_FAILSLAB
+    &failslab_attr.attr,
+#endif
+
+    NULL
+};
+
+static struct attribute_group slab_attr_group = {
+    .attrs = slab_attrs,
+};
+
+static ssize_t slab_attr_show(struct kobject *kobj,
+                struct attribute *attr,
+                char *buf)
+{
+    struct slab_attribute *attribute;
+    struct kmem_cache *s;
+    int err;
+
+    attribute = to_slab_attr(attr);
+    s = to_slab(kobj);
+
+    if (!attribute->show)
+        return -EIO;
+
+    err = attribute->show(s, buf);
+
+    return err;
+}
+
+static ssize_t slab_attr_store(struct kobject *kobj,
+                struct attribute *attr,
+                const char *buf, size_t len)
+{
+    struct slab_attribute *attribute;
+    struct kmem_cache *s;
+    int err;
+
+    attribute = to_slab_attr(attr);
+    s = to_slab(kobj);
+
+    if (!attribute->store)
+        return -EIO;
+
+    err = attribute->store(s, buf, len);
+#ifdef CONFIG_MEMCG_KMEM
+    if (slab_state >= FULL && err >= 0 && is_root_cache(s)) {
+        int i;
+
+        mutex_lock(&slab_mutex);
+        if (s->max_attr_size < len)
+            s->max_attr_size = len;
+
+        /*
+         * This is a best effort propagation, so this function's return
+         * value will be determined by the parent cache only. This is
+         * basically because not all attributes will have a well
+         * defined semantics for rollbacks - most of the actions will
+         * have permanent effects.
+         *
+         * Returning the error value of any of the children that fail
+         * is not 100 % defined, in the sense that users seeing the
+         * error code won't be able to know anything about the state of
+         * the cache.
+         *
+         * Only returning the error code for the parent cache at least
+         * has well defined semantics. The cache being written to
+         * directly either failed or succeeded, in which case we loop
+         * through the descendants with best-effort propagation.
+         */
+        for_each_memcg_cache_index(i) {
+            struct kmem_cache *c = cache_from_memcg(s, i);
+            if (c)
+                attribute->store(c, buf, len);
+        }
+        mutex_unlock(&slab_mutex);
+    }
+#endif
+    return err;
+}
+
+static void memcg_propagate_slab_attrs(struct kmem_cache *s)
+{
+#ifdef CONFIG_MEMCG_KMEM
+    int i;
+    char *buffer = NULL;
+
+    if (!is_root_cache(s))
+        return;
+
+    /*
+     * This mean this cache had no attribute written. Therefore, no point
+     * in copying default values around
+     */
+    if (!s->max_attr_size)
+        return;
+
+    for (i = 0; i < ARRAY_SIZE(slab_attrs); i++) {
+        char mbuf[64];
+        char *buf;
+        struct slab_attribute *attr = to_slab_attr(slab_attrs[i]);
+
+        if (!attr || !attr->store || !attr->show)
+            continue;
+
+        /*
+         * It is really bad that we have to allocate here, so we will
+         * do it only as a fallback. If we actually allocate, though,
+         * we can just use the allocated buffer until the end.
+         *
+         * Most of the slub attributes will tend to be very small in
+         * size, but sysfs allows buffers up to a page, so they can
+         * theoretically happen.
+         */
+        if (buffer)
+            buf = buffer;
+        else if (s->max_attr_size < ARRAY_SIZE(mbuf))
+            buf = mbuf;
+        else {
+            buffer = (char *) get_zeroed_page(GFP_KERNEL);
+            if (WARN_ON(!buffer))
+                continue;
+            buf = buffer;
+        }
+
+        attr->show(s->memcg_params->root_cache, buf);
+        attr->store(s, buf, strlen(buf));
+    }
+
+    if (buffer)
+        free_page((unsigned long)buffer);
+#endif
+}
+
+static const struct sysfs_ops slab_sysfs_ops = {
+    .show = slab_attr_show,
+    .store = slab_attr_store,
+};
+
+static struct kobj_type slab_ktype = {
+    .sysfs_ops = &slab_sysfs_ops,
+};
+
+static int uevent_filter(struct kset *kset, struct kobject *kobj)
+{
+    struct kobj_type *ktype = get_ktype(kobj);
+
+    if (ktype == &slab_ktype)
+        return 1;
+    return 0;
+}
+
+static const struct kset_uevent_ops slab_uevent_ops = {
+    .filter = uevent_filter,
+};
+
+static struct kset *slab_kset;
+
+#define ID_STR_LENGTH 64
+
+/* Create a unique string id for a slab cache:
+ *
+ * Format    :[flags-]size
+ */
+static char *create_unique_id(struct kmem_cache *s)
+{
+    char *name = kmalloc(ID_STR_LENGTH, GFP_KERNEL);
+    char *p = name;
+
+    BUG_ON(!name);
+
+    *p++ = ':';
+    /*
+     * First flags affecting slabcache operations. We will only
+     * get here for aliasable slabs so we do not need to support
+     * too many flags. The flags here must cover all flags that
+     * are matched during merging to guarantee that the id is
+     * unique.
+     */
+    if (s->flags & SLAB_CACHE_DMA)
+        *p++ = 'd';
+    if (s->flags & SLAB_RECLAIM_ACCOUNT)
+        *p++ = 'a';
+    if (s->flags & SLAB_DEBUG_FREE)
+        *p++ = 'F';
+    if (!(s->flags & SLAB_NOTRACK))
+        *p++ = 't';
+    if (p != name + 1)
+        *p++ = '-';
+    p += sprintf(p, "%07d", s->size);
+
+#ifdef CONFIG_MEMCG_KMEM
+    if (!is_root_cache(s))
+        p += sprintf(p, "-%08d", memcg_cache_id(s->memcg_params->memcg));
+#endif
+
+    BUG_ON(p > name + ID_STR_LENGTH - 1);
+    return name;
+}
+
+static int sysfs_slab_add(struct kmem_cache *s)
+{
+    int err;
+    const char *name;
+    int unmergeable = slab_unmergeable(s);
+
+    if (unmergeable) {
+        /*
+         * Slabcache can never be merged so we can use the name proper.
+         * This is typically the case for debug situations. In that
+         * case we can catch duplicate names easily.
+         */
+        sysfs_remove_link(&slab_kset->kobj, s->name);
+        name = s->name;
+    } else {
+        /*
+         * Create a unique name for the slab as a target
+         * for the symlinks.
+         */
+        name = create_unique_id(s);
+    }
+
+    s->kobj.kset = slab_kset;
+    err = kobject_init_and_add(&s->kobj, &slab_ktype, NULL, name);
+    if (err) {
+        kobject_put(&s->kobj);
+        return err;
+    }
+
+    err = sysfs_create_group(&s->kobj, &slab_attr_group);
+    if (err) {
+        kobject_del(&s->kobj);
+        kobject_put(&s->kobj);
+        return err;
+    }
+    kobject_uevent(&s->kobj, KOBJ_ADD);
+    if (!unmergeable) {
+        /* Setup first alias */
+        sysfs_slab_alias(s, s->name);
+        kfree(name);
+    }
+    return 0;
+}
+
+static void sysfs_slab_remove(struct kmem_cache *s)
+{
+    if (slab_state < FULL)
+        /*
+         * Sysfs has not been setup yet so no need to remove the
+         * cache from sysfs.
+         */
+        return;
+
+    kobject_uevent(&s->kobj, KOBJ_REMOVE);
+    kobject_del(&s->kobj);
+    kobject_put(&s->kobj);
+}
+
+/*
+ * Need to buffer aliases during bootup until sysfs becomes
+ * available lest we lose that information.
+ */
+struct saved_alias {
+    struct kmem_cache *s;
+    const char *name;
+    struct saved_alias *next;
+};
+
+static struct saved_alias *alias_list;
+
+static int sysfs_slab_alias(struct kmem_cache *s, const char *name)
+{
+    struct saved_alias *al;
+
+    if (slab_state == FULL) {
+        /*
+         * If we have a leftover link then remove it.
+         */
+        sysfs_remove_link(&slab_kset->kobj, name);
+        return sysfs_create_link(&slab_kset->kobj, &s->kobj, name);
+    }
+
+    al = kmalloc(sizeof(struct saved_alias), GFP_KERNEL);
+    if (!al)
+        return -ENOMEM;
+
+    al->s = s;
+    al->name = name;
+    al->next = alias_list;
+    alias_list = al;
+    return 0;
+}
+
+static int __init slab_sysfs_init(void)
+{
+    struct kmem_cache *s;
+    int err;
+
+    mutex_lock(&slab_mutex);
+
+    slab_kset = kset_create_and_add("slab", &slab_uevent_ops, kernel_kobj);
+    if (!slab_kset) {
+        mutex_unlock(&slab_mutex);
+        printk(KERN_ERR "Cannot register slab subsystem.\n");
+        return -ENOSYS;
+    }
+
+    slab_state = FULL;
+
+    list_for_each_entry(s, &slab_caches, list) {
+        err = sysfs_slab_add(s);
+        if (err)
+            printk(KERN_ERR "SLUB: Unable to add boot slab %s"
+                        " to sysfs\n", s->name);
+    }
+
+    while (alias_list) {
+        struct saved_alias *al = alias_list;
+
+        alias_list = alias_list->next;
+        err = sysfs_slab_alias(al->s, al->name);
+        if (err)
+            printk(KERN_ERR "SLUB: Unable to add boot slab alias"
+                    " %s to sysfs\n", al->name);
+        kfree(al);
+    }
+
+    mutex_unlock(&slab_mutex);
+    resiliency_test();
+    return 0;
+}
+
+__initcall(slab_sysfs_init);
+#endif /* CONFIG_SYSFS */
+
+/*
+ * The /proc/slabinfo ABI
+ */
+#ifdef CONFIG_SLABINFO
+void get_slabinfo(struct kmem_cache *s, struct slabinfo *sinfo)
+{
+    unsigned long nr_partials = 0;
+    unsigned long nr_slabs = 0;
+    unsigned long nr_objs = 0;
+    unsigned long nr_free = 0;
+    int node;
+
+    for_each_online_node(node) {
+        struct kmem_cache_node *n = get_node(s, node);
+
+        if (!n)
+            continue;
+
+        nr_partials += n->nr_partial;
+        nr_slabs += atomic_long_read(&n->nr_slabs);
+        nr_objs += atomic_long_read(&n->total_objects);
+        nr_free += count_partial(n, count_free);
+    }
+
+    sinfo->active_objs = nr_objs - nr_free;
+    sinfo->num_objs = nr_objs;
+    sinfo->active_slabs = nr_slabs;
+    sinfo->num_slabs = nr_slabs;
+    sinfo->objects_per_slab = oo_objects(s->oo);
+    sinfo->cache_order = oo_order(s->oo);
+}
+
+void slabinfo_show_stats(struct seq_file *m, struct kmem_cache *s)
+{
+}
+
+ssize_t slabinfo_write(struct file *file, const char __user *buffer,
+               size_t count, loff_t *ppos)
+{
+    return -EIO;
+}
+#endif /* CONFIG_SLABINFO */
diff --git a/redhat/configs/debug/CONFIG_SLEB_DEBUG_ON b/redhat/configs/debug/CONFIG_SLEB_DEBUG_ON
new file mode 100644
index 0000000..86debf1
--- /dev/null
+++ b/redhat/configs/debug/CONFIG_SLEB_DEBUG_ON
@@ -0,0 +1 @@
+# CONFIG_SLEB_DEBUG_ON is not set
diff --git a/redhat/configs/generic/CONFIG_SLEB b/redhat/configs/generic/CONFIG_SLEB
new file mode 100644
index 0000000..3a2f65d
--- /dev/null
+++ b/redhat/configs/generic/CONFIG_SLEB
@@ -0,0 +1 @@
+CONFIG_SLEB=y
diff --git a/redhat/configs/generic/CONFIG_SLEB_DEBUG b/redhat/configs/generic/CONFIG_SLEB_DEBUG
new file mode 100644
index 0000000..4bc4d4c
--- /dev/null
+++ b/redhat/configs/generic/CONFIG_SLEB_DEBUG
@@ -0,0 +1 @@
+CONFIG_SLEB_DEBUG=y
diff --git a/redhat/configs/generic/CONFIG_SLEB_DEBUG_ON b/redhat/configs/generic/CONFIG_SLEB_DEBUG_ON
new file mode 100644
index 0000000..86debf1
--- /dev/null
+++ b/redhat/configs/generic/CONFIG_SLEB_DEBUG_ON
@@ -0,0 +1 @@
+# CONFIG_SLEB_DEBUG_ON is not set
diff --git a/redhat/configs/generic/CONFIG_SLUB b/redhat/configs/generic/CONFIG_SLUB
index 05f729d..8023302 100644
--- a/redhat/configs/generic/CONFIG_SLUB
+++ b/redhat/configs/generic/CONFIG_SLUB
@@ -1 +1 @@
-CONFIG_SLUB=y
+# CONFIG_SLUB is not set
diff --git a/redhat/rhpkg/Makefile b/redhat/rhpkg/Makefile
index ebd7b00..e9983da 100644
--- a/redhat/rhpkg/Makefile
+++ b/redhat/rhpkg/Makefile
@@ -32,6 +32,7 @@ extremedebug:


 debug:
     @perl -pi -e 's/# CONFIG_SLUB_DEBUG_ON is not set/CONFIG_SLUB_DEBUG_ON=y/' config-nodebug
+    @perl -pi -e 's/# CONFIG_SLEB_DEBUG_ON is not set/CONFIG_SLEB_DEBUG_ON=y/' config-nodebug
     @perl -pi -e 's/# CONFIG_LOCK_STAT is not set/CONFIG_LOCK_STAT=y/' config-nodebug
     @perl -pi -e 's/# CONFIG_DEBUG_STACK_USAGE is not set/CONFIG_DEBUG_STACK_USAGE=y/' config-nodebug
     @perl -pi -e 's/# CONFIG_DEBUG_SLAB is not set/CONFIG_DEBUG_SLAB=y/' config-nodebug
@@ -109,6 +110,7 @@ nodebug: release
     @perl -pi -e 's/^%define debugbuildsenabled 1/%define debugbuildsenabled 0/' kernel.spec
 release:
     @perl -pi -e 's/CONFIG_SLUB_DEBUG_ON=y/# CONFIG_SLUB_DEBUG_ON is not set/' config-nodebug
+    @perl -pi -e 's/CONFIG_SLEB_DEBUG_ON=y/# CONFIG_SLEB_DEBUG_ON is not set/' config-nodebug
     @perl -pi -e 's/CONFIG_LOCK_STAT=y/# CONFIG_LOCK_STAT is not set/' config-nodebug
     @perl -pi -e 's/CONFIG_DEBUG_STACK_USAGE=y/# CONFIG_DEBUG_STACK_USAGE is not set/' config-nodebug
     @perl -pi -e 's/CONFIG_DEBUG_SLAB=y/# CONFIG_DEBUG_SLAB is not set/' config-nodebug
diff --git a/scripts/Makefile.modpost b/scripts/Makefile.modpost
index 8dcdca2..41051da 100644
--- a/scripts/Makefile.modpost
+++ b/scripts/Makefile.modpost
@@ -84,6 +84,13 @@ quiet_cmd_modpost = MODPOST $(words $(filter-out vmlinux FORCE, $^)) modules
       cmd_modpost = $(MODLISTCMD) | sed 's/\.ko$$/.o/' | $(modpost) -s -T -


 PHONY += __modpost
+__modpost_jgh: FORCE
+    <foo scripts/mod/modpost -m -a -o ./Module.symvers      -s -T - vmlinux
+
+__modpost_jgh2: FORCE
+    echo $(call cmd,modpost) $(wildcard vmlinux)
+    $(call cmd,modpost) $(wildcard vmlinux)
+
 __modpost: $(modules:.ko=.o) FORCE
     $(call cmd,modpost) $(wildcard vmlinux)