mm/zsmalloc: adjust order of functions
authorGanesh Mahendran <opensource.ganesh@gmail.com>
Fri, 19 Dec 2014 00:17:40 +0000 (16:17 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 19 Dec 2014 03:08:11 +0000 (19:08 -0800)
Currently functions in zsmalloc.c does not arranged in a readable and
reasonable sequence.  With the more and more functions added, we may
meet below inconvenience.  For example:

Current functions:

    void zs_init()
    {
    }

    static void get_maxobj_per_zspage()
    {
    }

Then I want to add a func_1() which is called from zs_init(), and this
new added function func_1() will used get_maxobj_per_zspage() which is
defined below zs_init().

    void func_1()
    {
        get_maxobj_per_zspage()
    }

    void zs_init()
    {
        func_1()
    }

    static void get_maxobj_per_zspage()
    {
    }

This will cause compiling issue. So we must add a declaration:

    static void get_maxobj_per_zspage();

before func_1() if we do not put get_maxobj_per_zspage() before
func_1().

In addition, puting module_[init|exit] functions at the bottom of the
file conforms to our habit.

So, this patch ajusts function sequence as:

    /* helper functions */
    ...
    obj_location_to_handle()
    ...

    /* Some exported functions */
    ...

    zs_map_object()
    zs_unmap_object()

    zs_malloc()
    zs_free()

    zs_init()
    zs_exit()

Signed-off-by: Ganesh Mahendran <opensource.ganesh@gmail.com>
Cc: Nitin Gupta <ngupta@vflare.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/zsmalloc.c

index 4d0a063145ec45647a65654dd09388e8b7936fde..b72403927aa4b06610cb08e9506d677c98a1a5cd 100644 (file)
@@ -884,19 +884,6 @@ static struct notifier_block zs_cpu_nb = {
        .notifier_call = zs_cpu_notifier
 };
 
-static void zs_unregister_cpu_notifier(void)
-{
-       int cpu;
-
-       cpu_notifier_register_begin();
-
-       for_each_online_cpu(cpu)
-               zs_cpu_notifier(NULL, CPU_DEAD, (void *)(long)cpu);
-       __unregister_cpu_notifier(&zs_cpu_nb);
-
-       cpu_notifier_register_done();
-}
-
 static int zs_register_cpu_notifier(void)
 {
        int cpu, uninitialized_var(ret);
@@ -914,40 +901,28 @@ static int zs_register_cpu_notifier(void)
        return notifier_to_errno(ret);
 }
 
-static void init_zs_size_classes(void)
+static void zs_unregister_cpu_notifier(void)
 {
-       int nr;
+       int cpu;
 
-       nr = (ZS_MAX_ALLOC_SIZE - ZS_MIN_ALLOC_SIZE) / ZS_SIZE_CLASS_DELTA + 1;
-       if ((ZS_MAX_ALLOC_SIZE - ZS_MIN_ALLOC_SIZE) % ZS_SIZE_CLASS_DELTA)
-               nr += 1;
+       cpu_notifier_register_begin();
 
-       zs_size_classes = nr;
-}
+       for_each_online_cpu(cpu)
+               zs_cpu_notifier(NULL, CPU_DEAD, (void *)(long)cpu);
+       __unregister_cpu_notifier(&zs_cpu_nb);
 
-static void __exit zs_exit(void)
-{
-#ifdef CONFIG_ZPOOL
-       zpool_unregister_driver(&zs_zpool_driver);
-#endif
-       zs_unregister_cpu_notifier();
+       cpu_notifier_register_done();
 }
 
-static int __init zs_init(void)
+static void init_zs_size_classes(void)
 {
-       int ret = zs_register_cpu_notifier();
-
-       if (ret) {
-               zs_unregister_cpu_notifier();
-               return ret;
-       }
+       int nr;
 
-       init_zs_size_classes();
+       nr = (ZS_MAX_ALLOC_SIZE - ZS_MIN_ALLOC_SIZE) / ZS_SIZE_CLASS_DELTA + 1;
+       if ((ZS_MAX_ALLOC_SIZE - ZS_MIN_ALLOC_SIZE) % ZS_SIZE_CLASS_DELTA)
+               nr += 1;
 
-#ifdef CONFIG_ZPOOL
-       zpool_register_driver(&zs_zpool_driver);
-#endif
-       return 0;
+       zs_size_classes = nr;
 }
 
 static unsigned int get_maxobj_per_zspage(int size, int pages_per_zspage)
@@ -967,113 +942,101 @@ static bool can_merge(struct size_class *prev, int size, int pages_per_zspage)
        return true;
 }
 
+unsigned long zs_get_total_pages(struct zs_pool *pool)
+{
+       return atomic_long_read(&pool->pages_allocated);
+}
+EXPORT_SYMBOL_GPL(zs_get_total_pages);
+
 /**
- * zs_create_pool - Creates an allocation pool to work from.
- * @flags: allocation flags used to allocate pool metadata
+ * zs_map_object - get address of allocated object from handle.
+ * @pool: pool from which the object was allocated
+ * @handle: handle returned from zs_malloc
  *
- * This function must be called before anything when using
- * the zsmalloc allocator.
+ * Before using an object allocated from zs_malloc, it must be mapped using
+ * this function. When done with the object, it must be unmapped using
+ * zs_unmap_object.
  *
- * On success, a pointer to the newly created pool is returned,
- * otherwise NULL.
+ * Only one object can be mapped per cpu at a time. There is no protection
+ * against nested mappings.
+ *
+ * This function returns with preemption and page faults disabled.
  */
-struct zs_pool *zs_create_pool(gfp_t flags)
+void *zs_map_object(struct zs_pool *pool, unsigned long handle,
+                       enum zs_mapmode mm)
 {
-       int i;
-       struct zs_pool *pool;
-       struct size_class *prev_class = NULL;
+       struct page *page;
+       unsigned long obj_idx, off;
 
-       pool = kzalloc(sizeof(*pool), GFP_KERNEL);
-       if (!pool)
-               return NULL;
+       unsigned int class_idx;
+       enum fullness_group fg;
+       struct size_class *class;
+       struct mapping_area *area;
+       struct page *pages[2];
 
-       pool->size_class = kcalloc(zs_size_classes, sizeof(struct size_class *),
-                       GFP_KERNEL);
-       if (!pool->size_class) {
-               kfree(pool);
-               return NULL;
-       }
+       BUG_ON(!handle);
 
        /*
-        * Iterate reversly, because, size of size_class that we want to use
-        * for merging should be larger or equal to current size.
+        * Because we use per-cpu mapping areas shared among the
+        * pools/users, we can't allow mapping in interrupt context
+        * because it can corrupt another users mappings.
         */
-       for (i = zs_size_classes - 1; i >= 0; i--) {
-               int size;
-               int pages_per_zspage;
-               struct size_class *class;
-
-               size = ZS_MIN_ALLOC_SIZE + i * ZS_SIZE_CLASS_DELTA;
-               if (size > ZS_MAX_ALLOC_SIZE)
-                       size = ZS_MAX_ALLOC_SIZE;
-               pages_per_zspage = get_pages_per_zspage(size);
-
-               /*
-                * size_class is used for normal zsmalloc operation such
-                * as alloc/free for that size. Although it is natural that we
-                * have one size_class for each size, there is a chance that we
-                * can get more memory utilization if we use one size_class for
-                * many different sizes whose size_class have same
-                * characteristics. So, we makes size_class point to
-                * previous size_class if possible.
-                */
-               if (prev_class) {
-                       if (can_merge(prev_class, size, pages_per_zspage)) {
-                               pool->size_class[i] = prev_class;
-                               continue;
-                       }
-               }
-
-               class = kzalloc(sizeof(struct size_class), GFP_KERNEL);
-               if (!class)
-                       goto err;
+       BUG_ON(in_interrupt());
 
-               class->size = size;
-               class->index = i;
-               class->pages_per_zspage = pages_per_zspage;
-               spin_lock_init(&class->lock);
-               pool->size_class[i] = class;
+       obj_handle_to_location(handle, &page, &obj_idx);
+       get_zspage_mapping(get_first_page(page), &class_idx, &fg);
+       class = pool->size_class[class_idx];
+       off = obj_idx_to_offset(page, obj_idx, class->size);
 
-               prev_class = class;
+       area = &get_cpu_var(zs_map_area);
+       area->vm_mm = mm;
+       if (off + class->size <= PAGE_SIZE) {
+               /* this object is contained entirely within a page */
+               area->vm_addr = kmap_atomic(page);
+               return area->vm_addr + off;
        }
 
-       pool->flags = flags;
-
-       return pool;
+       /* this object spans two pages */
+       pages[0] = page;
+       pages[1] = get_next_page(page);
+       BUG_ON(!pages[1]);
 
-err:
-       zs_destroy_pool(pool);
-       return NULL;
+       return __zs_map_object(area, pages, off, class->size);
 }
-EXPORT_SYMBOL_GPL(zs_create_pool);
+EXPORT_SYMBOL_GPL(zs_map_object);
 
-void zs_destroy_pool(struct zs_pool *pool)
+void zs_unmap_object(struct zs_pool *pool, unsigned long handle)
 {
-       int i;
+       struct page *page;
+       unsigned long obj_idx, off;
 
-       for (i = 0; i < zs_size_classes; i++) {
-               int fg;
-               struct size_class *class = pool->size_class[i];
+       unsigned int class_idx;
+       enum fullness_group fg;
+       struct size_class *class;
+       struct mapping_area *area;
 
-               if (!class)
-                       continue;
+       BUG_ON(!handle);
 
-               if (class->index != i)
-                       continue;
+       obj_handle_to_location(handle, &page, &obj_idx);
+       get_zspage_mapping(get_first_page(page), &class_idx, &fg);
+       class = pool->size_class[class_idx];
+       off = obj_idx_to_offset(page, obj_idx, class->size);
 
-               for (fg = 0; fg < _ZS_NR_FULLNESS_GROUPS; fg++) {
-                       if (class->fullness_list[fg]) {
-                               pr_info("Freeing non-empty class with size %db, fullness group %d\n",
-                                       class->size, fg);
-                       }
-               }
-               kfree(class);
-       }
+       area = this_cpu_ptr(&zs_map_area);
+       if (off + class->size <= PAGE_SIZE)
+               kunmap_atomic(area->vm_addr);
+       else {
+               struct page *pages[2];
 
-       kfree(pool->size_class);
-       kfree(pool);
+               pages[0] = page;
+               pages[1] = get_next_page(page);
+               BUG_ON(!pages[1]);
+
+               __zs_unmap_object(area, pages, off, class->size);
+       }
+       put_cpu_var(zs_map_area);
 }
-EXPORT_SYMBOL_GPL(zs_destroy_pool);
+EXPORT_SYMBOL_GPL(zs_unmap_object);
 
 /**
  * zs_malloc - Allocate block of given size from pool.
@@ -1176,100 +1139,137 @@ void zs_free(struct zs_pool *pool, unsigned long obj)
 EXPORT_SYMBOL_GPL(zs_free);
 
 /**
- * zs_map_object - get address of allocated object from handle.
- * @pool: pool from which the object was allocated
- * @handle: handle returned from zs_malloc
- *
- * Before using an object allocated from zs_malloc, it must be mapped using
- * this function. When done with the object, it must be unmapped using
- * zs_unmap_object.
+ * zs_create_pool - Creates an allocation pool to work from.
+ * @flags: allocation flags used to allocate pool metadata
  *
- * Only one object can be mapped per cpu at a time. There is no protection
- * against nested mappings.
+ * This function must be called before anything when using
+ * the zsmalloc allocator.
  *
- * This function returns with preemption and page faults disabled.
+ * On success, a pointer to the newly created pool is returned,
+ * otherwise NULL.
  */
-void *zs_map_object(struct zs_pool *pool, unsigned long handle,
-                       enum zs_mapmode mm)
+struct zs_pool *zs_create_pool(gfp_t flags)
 {
-       struct page *page;
-       unsigned long obj_idx, off;
+       int i;
+       struct zs_pool *pool;
+       struct size_class *prev_class = NULL;
 
-       unsigned int class_idx;
-       enum fullness_group fg;
-       struct size_class *class;
-       struct mapping_area *area;
-       struct page *pages[2];
+       pool = kzalloc(sizeof(*pool), GFP_KERNEL);
+       if (!pool)
+               return NULL;
 
-       BUG_ON(!handle);
+       pool->size_class = kcalloc(zs_size_classes, sizeof(struct size_class *),
+                       GFP_KERNEL);
+       if (!pool->size_class) {
+               kfree(pool);
+               return NULL;
+       }
 
        /*
-        * Because we use per-cpu mapping areas shared among the
-        * pools/users, we can't allow mapping in interrupt context
-        * because it can corrupt another users mappings.
+        * Iterate reversly, because, size of size_class that we want to use
+        * for merging should be larger or equal to current size.
         */
-       BUG_ON(in_interrupt());
+       for (i = zs_size_classes - 1; i >= 0; i--) {
+               int size;
+               int pages_per_zspage;
+               struct size_class *class;
 
-       obj_handle_to_location(handle, &page, &obj_idx);
-       get_zspage_mapping(get_first_page(page), &class_idx, &fg);
-       class = pool->size_class[class_idx];
-       off = obj_idx_to_offset(page, obj_idx, class->size);
+               size = ZS_MIN_ALLOC_SIZE + i * ZS_SIZE_CLASS_DELTA;
+               if (size > ZS_MAX_ALLOC_SIZE)
+                       size = ZS_MAX_ALLOC_SIZE;
+               pages_per_zspage = get_pages_per_zspage(size);
 
-       area = &get_cpu_var(zs_map_area);
-       area->vm_mm = mm;
-       if (off + class->size <= PAGE_SIZE) {
-               /* this object is contained entirely within a page */
-               area->vm_addr = kmap_atomic(page);
-               return area->vm_addr + off;
+               /*
+                * size_class is used for normal zsmalloc operation such
+                * as alloc/free for that size. Although it is natural that we
+                * have one size_class for each size, there is a chance that we
+                * can get more memory utilization if we use one size_class for
+                * many different sizes whose size_class have same
+                * characteristics. So, we makes size_class point to
+                * previous size_class if possible.
+                */
+               if (prev_class) {
+                       if (can_merge(prev_class, size, pages_per_zspage)) {
+                               pool->size_class[i] = prev_class;
+                               continue;
+                       }
+               }
+
+               class = kzalloc(sizeof(struct size_class), GFP_KERNEL);
+               if (!class)
+                       goto err;
+
+               class->size = size;
+               class->index = i;
+               class->pages_per_zspage = pages_per_zspage;
+               spin_lock_init(&class->lock);
+               pool->size_class[i] = class;
+
+               prev_class = class;
        }
 
-       /* this object spans two pages */
-       pages[0] = page;
-       pages[1] = get_next_page(page);
-       BUG_ON(!pages[1]);
+       pool->flags = flags;
 
-       return __zs_map_object(area, pages, off, class->size);
+       return pool;
+
+err:
+       zs_destroy_pool(pool);
+       return NULL;
 }
-EXPORT_SYMBOL_GPL(zs_map_object);
+EXPORT_SYMBOL_GPL(zs_create_pool);
 
-void zs_unmap_object(struct zs_pool *pool, unsigned long handle)
+void zs_destroy_pool(struct zs_pool *pool)
 {
-       struct page *page;
-       unsigned long obj_idx, off;
+       int i;
 
-       unsigned int class_idx;
-       enum fullness_group fg;
-       struct size_class *class;
-       struct mapping_area *area;
+       for (i = 0; i < zs_size_classes; i++) {
+               int fg;
+               struct size_class *class = pool->size_class[i];
 
-       BUG_ON(!handle);
+               if (!class)
+                       continue;
 
-       obj_handle_to_location(handle, &page, &obj_idx);
-       get_zspage_mapping(get_first_page(page), &class_idx, &fg);
-       class = pool->size_class[class_idx];
-       off = obj_idx_to_offset(page, obj_idx, class->size);
+               if (class->index != i)
+                       continue;
 
-       area = this_cpu_ptr(&zs_map_area);
-       if (off + class->size <= PAGE_SIZE)
-               kunmap_atomic(area->vm_addr);
-       else {
-               struct page *pages[2];
+               for (fg = 0; fg < _ZS_NR_FULLNESS_GROUPS; fg++) {
+                       if (class->fullness_list[fg]) {
+                               pr_info("Freeing non-empty class with size %db, fullness group %d\n",
+                                       class->size, fg);
+                       }
+               }
+               kfree(class);
+       }
 
-               pages[0] = page;
-               pages[1] = get_next_page(page);
-               BUG_ON(!pages[1]);
+       kfree(pool->size_class);
+       kfree(pool);
+}
+EXPORT_SYMBOL_GPL(zs_destroy_pool);
 
-               __zs_unmap_object(area, pages, off, class->size);
+static int __init zs_init(void)
+{
+       int ret = zs_register_cpu_notifier();
+
+       if (ret) {
+               zs_unregister_cpu_notifier();
+               return ret;
        }
-       put_cpu_var(zs_map_area);
+
+       init_zs_size_classes();
+
+#ifdef CONFIG_ZPOOL
+       zpool_register_driver(&zs_zpool_driver);
+#endif
+       return 0;
 }
-EXPORT_SYMBOL_GPL(zs_unmap_object);
 
-unsigned long zs_get_total_pages(struct zs_pool *pool)
+static void __exit zs_exit(void)
 {
-       return atomic_long_read(&pool->pages_allocated);
+#ifdef CONFIG_ZPOOL
+       zpool_unregister_driver(&zs_zpool_driver);
+#endif
+       zs_unregister_cpu_notifier();
 }
-EXPORT_SYMBOL_GPL(zs_get_total_pages);
 
 module_init(zs_init);
 module_exit(zs_exit);