drivers/misc/ll-pool.c - kernel/msm.git - Git at Google

 /*
  * Linked List Based Memory Allocator
  *
  * Copyright (C) 2019 Google, Inc.
  *
  * This software is licensed under the terms of the GNU General Public
  * License version 2, as published by the Free Software Foundation, and
  * may be copied, distributed, and modified under those terms.
  *
  * This program is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  * GNU General Public License for more details.
  */

 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

 /* Currently this allocator only supports management of memory up to 2G
  * size.
  */
 #define MAX_SIZE_SHIFT 32

 #include <linux/list.h>
 #include <linux/mm.h>
 #include <linux/mutex.h>
 #include <linux/ll-pool.h>
 #include <linux/slab.h>
 #include <linux/scatterlist.h>
 #include <linux/types.h>

 /**
  * ll_allocation: struct holding allocation information
  *
  * @nblks: number of blocks in this allocation
  * @size: total size allocated
  * @block_list: list of blocks that is allocated for this allocation
  * @table: scatter gather table for this allocation
  */
 struct ll_allocation {
 	size_t nblks;
 	size_t size;
 	struct list_head block_list;
 	struct sg_table table;
 };

 struct ll_mem_blk {
 	phys_addr_t offset;
 	size_t size;

 	struct list_head ll_head;
 };

 /* Caller must hold pool->lock */
 static void ll_free_blk(struct ll_pool *pool,
 		struct ll_mem_blk *blk)
 {
 	struct ll_mem_blk *fblk, *fblk_next, *fblk_prev;
 	struct list_head *fl_head;

 	fl_head = &pool->free_list_head;

 	list_del(&blk->ll_head);

 	/* Free list empty: add blk in */
 	if (list_empty(fl_head)) {
 		list_add_tail(&blk->ll_head, fl_head);
 		return;
 	}

 	/* If blk to free is before first free block, insert and merge if
 	 * possible
 	 */
 	fblk = list_first_entry(fl_head, struct ll_mem_blk, ll_head);
 	if (fblk->offset > blk->offset) {
 		if ((blk->offset + blk->size) == fblk->offset) {
 			fblk->size += blk->size;
 			fblk->offset -= blk->size;
 			kmem_cache_free(pool->mem_block_slab, blk);
 		} else {
 			list_add(&blk->ll_head, fl_head);
 		}

 		return;
 	}

 	/* if blk should be inserted at the end of the list:
 	 * This handles both general and singular list case
 	 */
 	fblk = list_last_entry(fl_head, struct ll_mem_blk, ll_head);
 	if (fblk->offset < blk->offset) {
 		/* merge block if possible */
 		if ((fblk->offset + fblk->size) == blk->offset) {
 			fblk->size += blk->size;
 			kmem_cache_free(pool->mem_block_slab, blk);
 		} else {
 			list_add_tail(&blk->ll_head, fl_head);
 		}

 		return;
 	}

 	/* In all other case, the free blok is to be inserted at the middle
 	 * of the list.
 	 */
 	fblk = list_first_entry(fl_head, struct ll_mem_blk, ll_head);
 	fblk = list_next_entry(fblk, ll_head);

 	list_for_each_entry_safe_from(fblk, fblk_next, fl_head, ll_head) {
 		if (blk->offset > fblk->offset)
 			continue;

 		fblk_prev = list_prev_entry(fblk, ll_head);
 		/* case 1: 3 block merge-able
 		 * case 2: if can merge with following block only
 		 * case 3: if can merge with previous block only
 		 * case 4: cannot merge
 		 */
 		if (((blk->offset + blk->size) == fblk->offset) &&
 				((fblk_prev->offset + fblk_prev->size) ==
 				 blk->offset)) {
 			fblk_prev->size += blk->size;
 			fblk_prev->size += fblk->size;

 			list_del_init(&fblk->ll_head);
 			kmem_cache_free(pool->mem_block_slab, fblk);
 			kmem_cache_free(pool->mem_block_slab, blk);
 		} else if ((blk->offset + blk->size) == fblk->offset) {
 			fblk->size += blk->size;
 			fblk->offset -= blk->size;
 			kmem_cache_free(pool->mem_block_slab, blk);
 		} else if ((fblk_prev->offset + fblk_prev->size) ==
 				blk->offset) {
 			fblk_prev->size += blk->size;
 			kmem_cache_free(pool->mem_block_slab, blk);
 		} else {
 			list_add(&blk->ll_head, &fblk_prev->ll_head);
 		}

 		return;
 	}
 }

 /* Caller must hold pool->lock */
 static void ll_free_allocation(struct ll_pool *pool,
 		struct ll_allocation *allocation)
 {
 	struct ll_mem_blk *blk, *blk_next;

 	/* For each allocation block, free and merge */
 	list_for_each_entry_safe(blk, blk_next, &allocation->block_list,
 			ll_head)
 		ll_free_blk(pool, blk);

 	/* Add freed memory size back to available size */
 	pool->remaining_size += allocation->size;

 	kmem_cache_free(pool->allocation_slab, allocation);
 }

 /* Caller must hold pool->lock */
 static struct ll_allocation *ll_alloc_contiguous(
 		struct ll_pool *pool, size_t len)
 {
 	struct ll_allocation *allocation;
 	struct ll_mem_blk *blk, *blk_next, *new_blk;

 	/* Creating allocation structure and initialize to contiguous */
 	allocation = kmem_cache_alloc(pool->allocation_slab, GFP_KERNEL);
 	if (!allocation)
 		return ERR_PTR(-ENOMEM);

 	allocation->nblks = 1;
 	allocation->size = len;
 	INIT_LIST_HEAD(&allocation->block_list);

 	list_for_each_entry_safe(blk, blk_next, &pool->free_list_head,
 			ll_head) {
 		if (blk->size < len)
 			continue;

 		if (blk->size == len) {
 			list_del_init(&blk->ll_head);
 			list_add_tail(&blk->ll_head, &allocation->block_list);
 		} else {
 			new_blk = kmem_cache_alloc(pool->mem_block_slab,
 					GFP_KERNEL);
 			if (!new_blk)
 				return ERR_PTR(-ENOMEM);

 			new_blk->size = len;
 			new_blk->offset = blk->offset;

 			list_add_tail(&new_blk->ll_head,
 					&allocation->block_list);

 			blk->size -= len;
 			blk->offset += len;
 		}

 		break;
 	}

 	/* If allocation failed return error */
 	if (list_empty(&allocation->block_list)) {
 		kmem_cache_free(pool->allocation_slab, allocation);
 		return ERR_PTR(-ENOMEM);
 	}

 	pool->remaining_size -= len;
 	return allocation;
 }

 /* Caller must hold pool->lock */
 static struct ll_allocation *ll_alloc_noncontiguous(
 		struct ll_pool *pool, size_t len)
 {
 	struct ll_allocation *allocation;
 	struct ll_mem_blk *blk, *blk_next, *new_blk;

 	/* Creating allocation structure and initialize to contiguous */
 	allocation = kmem_cache_alloc(pool->allocation_slab, GFP_KERNEL);
 	if (!allocation)
 		return ERR_PTR(-ENOMEM);

 	allocation->nblks = 0;
 	allocation->size = len;
 	INIT_LIST_HEAD(&allocation->block_list);

 	list_for_each_entry_safe(blk, blk_next, &pool->free_list_head,
 			ll_head) {
 		if (blk->size <= len) {
 			list_del(&blk->ll_head);
 			list_add_tail(&blk->ll_head, &allocation->block_list);
 			len -= blk->size;
 			allocation->nblks++;
 		} else {
 			new_blk = kmem_cache_alloc(pool->mem_block_slab,
 					GFP_KERNEL);
 			if (!new_blk)
 				return ERR_PTR(-ENOMEM);

 			new_blk->size = len;
 			new_blk->offset = blk->offset;
 			INIT_LIST_HEAD(&new_blk->ll_head);

 			list_add_tail(&new_blk->ll_head,
 					&allocation->block_list);
 			allocation->nblks++;

 			blk->size -= len;
 			blk->offset += len;

 			len = 0;
 		}

 		if (len == 0)
 			break;
 	}

 	if (len > 0) {
 		ll_free_allocation(pool, allocation);
 		pr_err("%s: possible inconsistent state or racing\n",
 				__func__);
 		return ERR_PTR(-ENOMEM);
 	}

 	pool->remaining_size -= allocation->size;
 	return allocation;
 }

 static int ll_build_sgtable(struct ll_pool *pool,
 		struct ll_allocation *allocation)
 {
 	struct ll_mem_blk *blk, *blk_next;
 	struct scatterlist *sg;
 	struct sg_table *table;
 	int ret;

 	if (unlikely(allocation->nblks == 0))
 		return -EINVAL;

 	table = &allocation->table;
 	ret = sg_alloc_table(table, allocation->nblks, GFP_KERNEL);
 	if (ret < 0)
 		return ret;

 	sg = table->sgl;
 	list_for_each_entry_safe(blk, blk_next, &allocation->block_list,
 			ll_head) {
 		sg_dma_address(sg) = pool->base_addr + blk->offset;
 		sg_dma_len(sg) = blk->size;
 		sg->length = blk->size;

 		sg = sg_next(sg);
 	}

 	return 0;
 }

 struct sg_table *ll_pool_alloc(struct ll_pool *pool, size_t len,
 		bool contiguous)
 {
 	int ret;
 	struct ll_allocation *alloc;

 	if (!pool) {
 		pr_err("%s: NULL ll pool\n", __func__);
 		return ERR_PTR(-EINVAL);
 	}

 	len = PAGE_ALIGN(len);

 	mutex_lock(&pool->lock);
 	if (len > pool->remaining_size || len == 0) {
 		pr_err("%s: Unable to allocate %llu bytes\n", __func__,
 				len);
 		mutex_unlock(&pool->lock);
 		return ERR_PTR(-ENOMEM);
 	}

 	if (contiguous)
 		alloc = ll_alloc_contiguous(pool, len);
 	else
 		alloc = ll_alloc_noncontiguous(pool, len);

 	mutex_unlock(&pool->lock);

 	if (IS_ERR(alloc))
 		return (void *)alloc;

 	ret = ll_build_sgtable(pool, alloc);
 	if (ret < 0) {
 		mutex_lock(&pool->lock);
 		ll_free_allocation(pool, alloc);
 		mutex_unlock(&pool->lock);

 		return ERR_PTR(ret);
 	}

 	return &alloc->table;
 }

 void ll_pool_free(struct ll_pool *pool, struct sg_table *table)
 {
 	struct ll_allocation *alloc;

 	if (!table)
 		return;

 	alloc = container_of(table, struct ll_allocation, table);
 	sg_free_table(table);

 	mutex_lock(&pool->lock);
 	ll_free_allocation(pool, alloc);
 	mutex_unlock(&pool->lock);
 }

 struct ll_pool *ll_pool_create(phys_addr_t base, size_t size)
 {
 	int ret;
 	struct ll_pool *pool;
 	struct ll_mem_blk *blk;

 	/* check if base is aligned with minimal order */
 	if (!IS_ALIGNED(base, PAGE_SIZE)) {
 		pr_err("%s: base address:0x%016lx is not page aligned\n",
 				__func__, base);
 		return ERR_PTR(-EINVAL);
 	}

 	/* check if size is aligned with minimal order and is power of 2 */
 	if (!IS_ALIGNED(size, PAGE_SIZE)) {
 		pr_err("%s: size: 0x%016lx is not page aligned and/or is not power of power of 2\n",
 				__func__, size);
 		return ERR_PTR(-EINVAL);
 	}

 	pool = kzalloc(sizeof(*pool), GFP_KERNEL);
 	if (!pool)
 		return ERR_PTR(-ENOMEM);

 	pool->base_addr = base;
 	pool->remaining_size = size;

 	mutex_init(&pool->lock);
 	INIT_LIST_HEAD(&pool->free_list_head);

 	/* Creating slab for mem block object */
 	pool->mem_block_slab = kmem_cache_create("ll_mem_block",
 			sizeof(struct ll_mem_blk), 0,
 			/* flag */ 0, /* ctor */ NULL);
 	if (!pool->mem_block_slab) {
 		pr_err("%s: creating mem block slab failed\n", __func__);
 		ret = -ENOMEM;
 		goto free_pool;
 	}

 	/* Creating slab for allocation object */
 	pool->allocation_slab = kmem_cache_create("ll_allocation",
 			sizeof(struct ll_allocation), 0,
 			/* flag */ 0, /* ctor */ NULL);
 	if (!pool->allocation_slab) {
 		pr_err("%s: creating allocation object slab failed\n",
 				__func__);
 		ret = -ENOMEM;
 		goto destroy_mem_block_slab;
 	}

 	/* Allocate the mem block that represent the full range */
 	blk = kmem_cache_alloc(pool->mem_block_slab, GFP_KERNEL);
 	if (!blk) {
 		pr_err("%s: allocating mem block from slab failed\n",
 				__func__);
 		ret = -ENOMEM;
 		goto destroy_allocation_slab;
 	}

 	blk->offset = 0;
 	blk->size = size;

 	list_add_tail(&blk->ll_head, &pool->free_list_head);

 	return pool;

 destroy_allocation_slab:
 	kmem_cache_destroy(pool->allocation_slab);
 destroy_mem_block_slab:
 	kmem_cache_destroy(pool->mem_block_slab);
 free_pool:
 	kfree(pool);
 	return ERR_PTR(ret);
 }

 static void ll_free_all_free_mem_blocks_struct(
 		struct ll_pool *pool)
 {
 	struct ll_mem_blk *blk, *blk_next;

 	list_for_each_entry_safe(blk, blk_next, &pool->free_list_head,
 			ll_head)
 		kmem_cache_free(pool->mem_block_slab, blk);
 }

 void ll_pool_destroy(struct ll_pool *pool)
 {
 	if (!pool)
 		return;

 	/* Warning if allocations still exist */
 	WARN_ON(!list_is_singular(&pool->free_list_head));

 	if (!list_empty(&pool->free_list_head))
 		ll_free_all_free_mem_blocks_struct(pool);

 	kmem_cache_destroy(pool->mem_block_slab);
 	kmem_cache_destroy(pool->allocation_slab);
 	kfree(pool);
 }
	/*
	* Linked List Based Memory Allocator
	*
	* Copyright (C) 2019 Google, Inc.
	*
	* This software is licensed under the terms of the GNU General Public
	* License version 2, as published by the Free Software Foundation, and
	* may be copied, distributed, and modified under those terms.
	*
	* This program is distributed in the hope that it will be useful,
	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	* GNU General Public License for more details.
	*/

	#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

	/* Currently this allocator only supports management of memory up to 2G
	* size.
	*/
	#define MAX_SIZE_SHIFT 32

	#include <linux/list.h>
	#include <linux/mm.h>
	#include <linux/mutex.h>
	#include <linux/ll-pool.h>
	#include <linux/slab.h>
	#include <linux/scatterlist.h>
	#include <linux/types.h>

	/**
	* ll_allocation: struct holding allocation information
	*
	* @nblks: number of blocks in this allocation
	* @size: total size allocated
	* @block_list: list of blocks that is allocated for this allocation
	* @table: scatter gather table for this allocation
	*/
	struct ll_allocation {
	size_t nblks;
	size_t size;
	struct list_head block_list;
	struct sg_table table;
	};

	struct ll_mem_blk {
	phys_addr_t offset;
	size_t size;

	struct list_head ll_head;
	};

	/* Caller must hold pool->lock */
	static void ll_free_blk(struct ll_pool *pool,
	struct ll_mem_blk *blk)
	{
	struct ll_mem_blk fblk, fblk_next, *fblk_prev;
	struct list_head *fl_head;

	fl_head = &pool->free_list_head;

	list_del(&blk->ll_head);

	/* Free list empty: add blk in */
	if (list_empty(fl_head)) {
	list_add_tail(&blk->ll_head, fl_head);
	return;
	}

	/* If blk to free is before first free block, insert and merge if
	* possible
	*/
	fblk = list_first_entry(fl_head, struct ll_mem_blk, ll_head);
	if (fblk->offset > blk->offset) {
	if ((blk->offset + blk->size) == fblk->offset) {
	fblk->size += blk->size;
	fblk->offset -= blk->size;
	kmem_cache_free(pool->mem_block_slab, blk);
	} else {
	list_add(&blk->ll_head, fl_head);
	}

	return;
	}

	/* if blk should be inserted at the end of the list:
	* This handles both general and singular list case
	*/
	fblk = list_last_entry(fl_head, struct ll_mem_blk, ll_head);
	if (fblk->offset < blk->offset) {
	/* merge block if possible */
	if ((fblk->offset + fblk->size) == blk->offset) {
	fblk->size += blk->size;
	kmem_cache_free(pool->mem_block_slab, blk);
	} else {
	list_add_tail(&blk->ll_head, fl_head);
	}

	return;
	}

	/* In all other case, the free blok is to be inserted at the middle
	* of the list.
	*/
	fblk = list_first_entry(fl_head, struct ll_mem_blk, ll_head);
	fblk = list_next_entry(fblk, ll_head);

	list_for_each_entry_safe_from(fblk, fblk_next, fl_head, ll_head) {
	if (blk->offset > fblk->offset)
	continue;

	fblk_prev = list_prev_entry(fblk, ll_head);
	/* case 1: 3 block merge-able
	* case 2: if can merge with following block only
	* case 3: if can merge with previous block only
	* case 4: cannot merge
	*/
	if (((blk->offset + blk->size) == fblk->offset) &&
	((fblk_prev->offset + fblk_prev->size) ==
	blk->offset)) {
	fblk_prev->size += blk->size;
	fblk_prev->size += fblk->size;

	list_del_init(&fblk->ll_head);
	kmem_cache_free(pool->mem_block_slab, fblk);
	kmem_cache_free(pool->mem_block_slab, blk);
	} else if ((blk->offset + blk->size) == fblk->offset) {
	fblk->size += blk->size;
	fblk->offset -= blk->size;
	kmem_cache_free(pool->mem_block_slab, blk);
	} else if ((fblk_prev->offset + fblk_prev->size) ==
	blk->offset) {
	fblk_prev->size += blk->size;
	kmem_cache_free(pool->mem_block_slab, blk);
	} else {
	list_add(&blk->ll_head, &fblk_prev->ll_head);
	}

	return;
	}
	}

	/* Caller must hold pool->lock */
	static void ll_free_allocation(struct ll_pool *pool,
	struct ll_allocation *allocation)
	{
	struct ll_mem_blk blk, blk_next;

	/* For each allocation block, free and merge */
	list_for_each_entry_safe(blk, blk_next, &allocation->block_list,
	ll_head)
	ll_free_blk(pool, blk);

	/* Add freed memory size back to available size */
	pool->remaining_size += allocation->size;

	kmem_cache_free(pool->allocation_slab, allocation);
	}

	/* Caller must hold pool->lock */
	static struct ll_allocation *ll_alloc_contiguous(
	struct ll_pool *pool, size_t len)
	{
	struct ll_allocation *allocation;
	struct ll_mem_blk blk, blk_next, *new_blk;

	/* Creating allocation structure and initialize to contiguous */
	allocation = kmem_cache_alloc(pool->allocation_slab, GFP_KERNEL);
	if (!allocation)
	return ERR_PTR(-ENOMEM);

	allocation->nblks = 1;
	allocation->size = len;
	INIT_LIST_HEAD(&allocation->block_list);

	list_for_each_entry_safe(blk, blk_next, &pool->free_list_head,
	ll_head) {
	if (blk->size < len)
	continue;

	if (blk->size == len) {
	list_del_init(&blk->ll_head);
	list_add_tail(&blk->ll_head, &allocation->block_list);
	} else {
	new_blk = kmem_cache_alloc(pool->mem_block_slab,
	GFP_KERNEL);
	if (!new_blk)
	return ERR_PTR(-ENOMEM);

	new_blk->size = len;
	new_blk->offset = blk->offset;

	list_add_tail(&new_blk->ll_head,
	&allocation->block_list);

	blk->size -= len;
	blk->offset += len;
	}

	break;
	}

	/* If allocation failed return error */
	if (list_empty(&allocation->block_list)) {
	kmem_cache_free(pool->allocation_slab, allocation);
	return ERR_PTR(-ENOMEM);
	}

	pool->remaining_size -= len;
	return allocation;
	}

	/* Caller must hold pool->lock */
	static struct ll_allocation *ll_alloc_noncontiguous(
	struct ll_pool *pool, size_t len)
	{
	struct ll_allocation *allocation;
	struct ll_mem_blk blk, blk_next, *new_blk;

	/* Creating allocation structure and initialize to contiguous */
	allocation = kmem_cache_alloc(pool->allocation_slab, GFP_KERNEL);
	if (!allocation)
	return ERR_PTR(-ENOMEM);

	allocation->nblks = 0;
	allocation->size = len;
	INIT_LIST_HEAD(&allocation->block_list);

	list_for_each_entry_safe(blk, blk_next, &pool->free_list_head,
	ll_head) {
	if (blk->size <= len) {
	list_del(&blk->ll_head);
	list_add_tail(&blk->ll_head, &allocation->block_list);
	len -= blk->size;
	allocation->nblks++;
	} else {
	new_blk = kmem_cache_alloc(pool->mem_block_slab,
	GFP_KERNEL);
	if (!new_blk)
	return ERR_PTR(-ENOMEM);

	new_blk->size = len;
	new_blk->offset = blk->offset;
	INIT_LIST_HEAD(&new_blk->ll_head);

	list_add_tail(&new_blk->ll_head,
	&allocation->block_list);
	allocation->nblks++;

	blk->size -= len;
	blk->offset += len;

	len = 0;
	}

	if (len == 0)
	break;
	}

	if (len > 0) {
	ll_free_allocation(pool, allocation);
	pr_err("%s: possible inconsistent state or racing\n",
	__func__);
	return ERR_PTR(-ENOMEM);
	}

	pool->remaining_size -= allocation->size;
	return allocation;
	}

	static int ll_build_sgtable(struct ll_pool *pool,
	struct ll_allocation *allocation)
	{
	struct ll_mem_blk blk, blk_next;
	struct scatterlist *sg;
	struct sg_table *table;
	int ret;

	if (unlikely(allocation->nblks == 0))
	return -EINVAL;

	table = &allocation->table;
	ret = sg_alloc_table(table, allocation->nblks, GFP_KERNEL);
	if (ret < 0)
	return ret;

	sg = table->sgl;
	list_for_each_entry_safe(blk, blk_next, &allocation->block_list,
	ll_head) {
	sg_dma_address(sg) = pool->base_addr + blk->offset;
	sg_dma_len(sg) = blk->size;
	sg->length = blk->size;

	sg = sg_next(sg);
	}

	return 0;
	}

	struct sg_table ll_pool_alloc(struct ll_pool pool, size_t len,
	bool contiguous)
	{
	int ret;
	struct ll_allocation *alloc;

	if (!pool) {
	pr_err("%s: NULL ll pool\n", __func__);
	return ERR_PTR(-EINVAL);
	}

	len = PAGE_ALIGN(len);

	mutex_lock(&pool->lock);
	if (len > pool->remaining_size \|\| len == 0) {
	pr_err("%s: Unable to allocate %llu bytes\n", __func__,
	len);
	mutex_unlock(&pool->lock);
	return ERR_PTR(-ENOMEM);
	}

	if (contiguous)
	alloc = ll_alloc_contiguous(pool, len);
	else
	alloc = ll_alloc_noncontiguous(pool, len);

	mutex_unlock(&pool->lock);

	if (IS_ERR(alloc))
	return (void *)alloc;

	ret = ll_build_sgtable(pool, alloc);
	if (ret < 0) {
	mutex_lock(&pool->lock);
	ll_free_allocation(pool, alloc);
	mutex_unlock(&pool->lock);

	return ERR_PTR(ret);
	}

	return &alloc->table;
	}

	void ll_pool_free(struct ll_pool pool, struct sg_table table)
	{
	struct ll_allocation *alloc;

	if (!table)
	return;

	alloc = container_of(table, struct ll_allocation, table);
	sg_free_table(table);

	mutex_lock(&pool->lock);
	ll_free_allocation(pool, alloc);
	mutex_unlock(&pool->lock);
	}

	struct ll_pool *ll_pool_create(phys_addr_t base, size_t size)
	{
	int ret;
	struct ll_pool *pool;
	struct ll_mem_blk *blk;

	/* check if base is aligned with minimal order */
	if (!IS_ALIGNED(base, PAGE_SIZE)) {
	pr_err("%s: base address:0x%016lx is not page aligned\n",
	__func__, base);
	return ERR_PTR(-EINVAL);
	}

	/* check if size is aligned with minimal order and is power of 2 */
	if (!IS_ALIGNED(size, PAGE_SIZE)) {
	pr_err("%s: size: 0x%016lx is not page aligned and/or is not power of power of 2\n",
	__func__, size);
	return ERR_PTR(-EINVAL);
	}

	pool = kzalloc(sizeof(*pool), GFP_KERNEL);
	if (!pool)
	return ERR_PTR(-ENOMEM);

	pool->base_addr = base;
	pool->remaining_size = size;

	mutex_init(&pool->lock);
	INIT_LIST_HEAD(&pool->free_list_head);

	/* Creating slab for mem block object */
	pool->mem_block_slab = kmem_cache_create("ll_mem_block",
	sizeof(struct ll_mem_blk), 0,
	/* flag / 0, / ctor */ NULL);
	if (!pool->mem_block_slab) {
	pr_err("%s: creating mem block slab failed\n", __func__);
	ret = -ENOMEM;
	goto free_pool;
	}

	/* Creating slab for allocation object */
	pool->allocation_slab = kmem_cache_create("ll_allocation",
	sizeof(struct ll_allocation), 0,
	/* flag / 0, / ctor */ NULL);
	if (!pool->allocation_slab) {
	pr_err("%s: creating allocation object slab failed\n",
	__func__);
	ret = -ENOMEM;
	goto destroy_mem_block_slab;
	}

	/* Allocate the mem block that represent the full range */
	blk = kmem_cache_alloc(pool->mem_block_slab, GFP_KERNEL);
	if (!blk) {
	pr_err("%s: allocating mem block from slab failed\n",
	__func__);
	ret = -ENOMEM;
	goto destroy_allocation_slab;
	}

	blk->offset = 0;
	blk->size = size;

	list_add_tail(&blk->ll_head, &pool->free_list_head);

	return pool;

	destroy_allocation_slab:
	kmem_cache_destroy(pool->allocation_slab);
	destroy_mem_block_slab:
	kmem_cache_destroy(pool->mem_block_slab);
	free_pool:
	kfree(pool);
	return ERR_PTR(ret);
	}

	static void ll_free_all_free_mem_blocks_struct(
	struct ll_pool *pool)
	{
	struct ll_mem_blk blk, blk_next;

	list_for_each_entry_safe(blk, blk_next, &pool->free_list_head,
	ll_head)
	kmem_cache_free(pool->mem_block_slab, blk);
	}

	void ll_pool_destroy(struct ll_pool *pool)
	{
	if (!pool)
	return;

	/* Warning if allocations still exist */
	WARN_ON(!list_is_singular(&pool->free_list_head));

	if (!list_empty(&pool->free_list_head))
	ll_free_all_free_mem_blocks_struct(pool);

	kmem_cache_destroy(pool->mem_block_slab);
	kmem_cache_destroy(pool->allocation_slab);
	kfree(pool);
	}