[PATCH] change gen_pool allocator to not touch managed memory

Modify the gen_pool allocator (lib/genalloc.c) to utilize a bitmap scheme
instead of the buddy scheme.  The purpose of this change is to eliminate
the touching of the actual memory being allocated.

Since the change modifies the interface, a change to the uncached allocator
(arch/ia64/kernel/uncached.c) is also required.

Both Andrey Volkov and Jes Sorenson have expressed a desire that the
gen_pool allocator not write to the memory being managed. See the
following:

  http://marc.theaimsgroup.com/?l=linux-kernel&m=113518602713125&w=2
  http://marc.theaimsgroup.com/?l=linux-kernel&m=113533568827916&w=2

Signed-off-by: Dean Nelson <dcn@sgi.com>
Cc: Andrey Volkov <avolkov@varma-el.com>
Acked-by: Jes Sorensen <jes@trained-monkey.org>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

arch/ia64/kernel/uncached.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2001-2005 Silicon Graphics, Inc.  All rights reserved.
+ * Copyright (C) 2001-2006 Silicon Graphics, Inc.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms of version 2 of the GNU General Public License
@@ -29,15 +29,8 @@
 #include <asm/tlbflush.h>
 #include <asm/sn/arch.h>
 
-#define DEBUG	0
 
-#if DEBUG
+extern void __init efi_memmap_walk_uc(efi_freemem_callback_t, void *);
-#define dprintk			printk
-#else
-#define dprintk(x...)		do { } while (0)
-#endif
-
-void __init efi_memmap_walk_uc (efi_freemem_callback_t callback);
 
 #define MAX_UNCACHED_GRANULES	5
 static int allocated_granules;
@@ -60,6 +53,7 @@ static void uncached_ipi_visibility(void *data)
 static void uncached_ipi_mc_drain(void *data)
 {
 	int status;
+
 	status = ia64_pal_mc_drain();
 	if (status)
 		printk(KERN_WARNING "ia64_pal_mc_drain() failed with %i on "
@@ -67,30 +61,35 @@ static void uncached_ipi_mc_drain(void *data)
 }
 
 
-static unsigned long
+/*
-uncached_get_new_chunk(struct gen_pool *poolp)
+ * Add a new chunk of uncached memory pages to the specified pool.
+ *
+ * @pool: pool to add new chunk of uncached memory to
+ * @nid: node id of node to allocate memory from, or -1
+ *
+ * This is accomplished by first allocating a granule of cached memory pages
+ * and then converting them to uncached memory pages.
+ */
+static int uncached_add_chunk(struct gen_pool *pool, int nid)
 {
 	struct page *page;
-	void *tmp;
 	int status, i;
-	unsigned long addr, node;
+	unsigned long c_addr, uc_addr;
 
 	if (allocated_granules >= MAX_UNCACHED_GRANULES)
-		return 0;
+		return -1;
 
-	node = poolp->private;
+	/* attempt to allocate a granule's worth of cached memory pages */
-	page = alloc_pages_node(node, GFP_KERNEL | __GFP_ZERO,
+
+	page = alloc_pages_node(nid, GFP_KERNEL | __GFP_ZERO,
 				IA64_GRANULE_SHIFT-PAGE_SHIFT);
-
-	dprintk(KERN_INFO "get_new_chunk page %p, addr %lx\n",
-		page, (unsigned long)(page-vmem_map) << PAGE_SHIFT);
-
-	/*
-	 * Do magic if no mem on local node! XXX
-	 */
 	if (!page)
-		return 0;
+		return -1;
-	tmp = page_address(page);
+
+	/* convert the memory pages from cached to uncached */
+
+	c_addr = (unsigned long)page_address(page);
+	uc_addr = c_addr - PAGE_OFFSET + __IA64_UNCACHED_OFFSET;
 
 	/*
 	 * There's a small race here where it's possible for someone to
@@ -100,76 +99,90 @@ uncached_get_new_chunk(struct gen_pool *poolp)
 	for (i = 0; i < (IA64_GRANULE_SIZE / PAGE_SIZE); i++)
 		SetPageUncached(&page[i]);
 
-	flush_tlb_kernel_range(tmp, tmp + IA64_GRANULE_SIZE);
+	flush_tlb_kernel_range(uc_addr, uc_adddr + IA64_GRANULE_SIZE);
 
 	status = ia64_pal_prefetch_visibility(PAL_VISIBILITY_PHYSICAL);
-
-	dprintk(KERN_INFO "pal_prefetch_visibility() returns %i on cpu %i\n",
-		status, raw_smp_processor_id());
-
 	if (!status) {
 		status = smp_call_function(uncached_ipi_visibility, NULL, 0, 1);
 		if (status)
-			printk(KERN_WARNING "smp_call_function failed for "
+			goto failed;
-			       "uncached_ipi_visibility! (%i)\n", status);
 	}
 
+	preempt_disable();
+
 	if (ia64_platform_is("sn2"))
-		sn_flush_all_caches((unsigned long)tmp, IA64_GRANULE_SIZE);
+		sn_flush_all_caches(uc_addr, IA64_GRANULE_SIZE);
 	else
-		flush_icache_range((unsigned long)tmp,
+		flush_icache_range(uc_addr, uc_addr + IA64_GRANULE_SIZE);
-				   (unsigned long)tmp+IA64_GRANULE_SIZE);
+
+	/* flush the just introduced uncached translation from the TLB */
+	local_flush_tlb_all();
+
+	preempt_enable();
 
 	ia64_pal_mc_drain();
 	status = smp_call_function(uncached_ipi_mc_drain, NULL, 0, 1);
 	if (status)
-		printk(KERN_WARNING "smp_call_function failed for "
+		goto failed;
-		       "uncached_ipi_mc_drain! (%i)\n", status);
 
-	addr = (unsigned long)tmp - PAGE_OFFSET + __IA64_UNCACHED_OFFSET;
+	/*
+	 * The chunk of memory pages has been converted to uncached so now we
+	 * can add it to the pool.
+	 */
+	status = gen_pool_add(pool, uc_addr, IA64_GRANULE_SIZE, nid);
+	if (status)
+		goto failed;
 
 	allocated_granules++;
-	return addr;
+	return 0;
+
+	/* failed to convert or add the chunk so give it back to the kernel */
+failed:
+	for (i = 0; i < (IA64_GRANULE_SIZE / PAGE_SIZE); i++)
+		ClearPageUncached(&page[i]);
+
+	free_pages(c_addr, IA64_GRANULE_SHIFT-PAGE_SHIFT);
+	return -1;
 }
 
 
 /*
  * uncached_alloc_page
  *
+ * @starting_nid: node id of node to start with, or -1
+ *
  * Allocate 1 uncached page. Allocates on the requested node. If no
  * uncached pages are available on the requested node, roundrobin starting
- * with higher nodes.
+ * with the next higher node.
  */
-unsigned long
+unsigned long uncached_alloc_page(int starting_nid)
-uncached_alloc_page(int nid)
 {
-	unsigned long maddr;
+	unsigned long uc_addr;
+	struct gen_pool *pool;
+	int nid;
 
-	maddr = gen_pool_alloc(uncached_pool[nid], PAGE_SIZE);
+	if (unlikely(starting_nid >= MAX_NUMNODES))
+		return 0;
 
-	dprintk(KERN_DEBUG "uncached_alloc_page returns %lx on node %i\n",
+	if (starting_nid < 0)
-		maddr, nid);
+		starting_nid = numa_node_id();
+	nid = starting_nid;
 
-	/*
+	do {
-	 * If no memory is availble on our local node, try the
+		if (!node_online(nid))
-	 * remaining nodes in the system.
-	 */
-	if (!maddr) {
-		int i;
-
-		for (i = MAX_NUMNODES - 1; i >= 0; i--) {
-			if (i == nid || !node_online(i))
 			continue;
-			maddr = gen_pool_alloc(uncached_pool[i], PAGE_SIZE);
+		pool = uncached_pool[nid];
-			dprintk(KERN_DEBUG "uncached_alloc_page alternate search "
+		if (pool == NULL)
-				"returns %lx on node %i\n", maddr, i);
+			continue;
-			if (maddr) {
+		do {
-				break;
+			uc_addr = gen_pool_alloc(pool, PAGE_SIZE);
-			}
+			if (uc_addr != 0)
-		}
+				return uc_addr;
-	}
+		} while (uncached_add_chunk(pool, nid) == 0);
 
-	return maddr;
+	} while ((nid = (nid + 1) % MAX_NUMNODES) != starting_nid);
+
+	return 0;
 }
 EXPORT_SYMBOL(uncached_alloc_page);
 
@@ -177,21 +190,22 @@ EXPORT_SYMBOL(uncached_alloc_page);
 /*
  * uncached_free_page
  *
+ * @uc_addr: uncached address of page to free
+ *
  * Free a single uncached page.
  */
-void
+void uncached_free_page(unsigned long uc_addr)
-uncached_free_page(unsigned long maddr)
 {
-	int node;
+	int nid = paddr_to_nid(uc_addr - __IA64_UNCACHED_OFFSET);
+	struct gen_pool *pool = uncached_pool[nid];
 
-	node = paddr_to_nid(maddr - __IA64_UNCACHED_OFFSET);
+	if (unlikely(pool == NULL))
+		return;
 
-	dprintk(KERN_DEBUG "uncached_free_page(%lx) on node %i\n", maddr, node);
+	if ((uc_addr & (0XFUL << 60)) != __IA64_UNCACHED_OFFSET)
+		panic("uncached_free_page invalid address %lx\n", uc_addr);
 
-	if ((maddr & (0XFUL << 60)) != __IA64_UNCACHED_OFFSET)
+	gen_pool_free(pool, uc_addr, PAGE_SIZE);
-		panic("uncached_free_page invalid address %lx\n", maddr);
-
-	gen_pool_free(uncached_pool[node], maddr, PAGE_SIZE);
 }
 EXPORT_SYMBOL(uncached_free_page);
 
@@ -199,43 +213,39 @@ EXPORT_SYMBOL(uncached_free_page);
 /*
  * uncached_build_memmap,
  *
+ * @uc_start: uncached starting address of a chunk of uncached memory
+ * @uc_end: uncached ending address of a chunk of uncached memory
+ * @arg: ignored, (NULL argument passed in on call to efi_memmap_walk_uc())
+ *
  * Called at boot time to build a map of pages that can be used for
  * memory special operations.
  */
-static int __init
+static int __init uncached_build_memmap(unsigned long uc_start,
-uncached_build_memmap(unsigned long start, unsigned long end, void *arg)
+					unsigned long uc_end, void *arg)
 {
-	long length = end - start;
+	int nid = paddr_to_nid(uc_start - __IA64_UNCACHED_OFFSET);
-	int node;
+	struct gen_pool *pool = uncached_pool[nid];
-
+	size_t size = uc_end - uc_start;
-	dprintk(KERN_ERR "uncached_build_memmap(%lx %lx)\n", start, end);
 
 	touch_softlockup_watchdog();
-	memset((char *)start, 0, length);
 
-	node = paddr_to_nid(start - __IA64_UNCACHED_OFFSET);
+	if (pool != NULL) {
-
+		memset((char *)uc_start, 0, size);
-	for (; start < end ; start += PAGE_SIZE) {
+		(void) gen_pool_add(pool, uc_start, size, nid);
-		dprintk(KERN_INFO "sticking %lx into the pool!\n", start);
-		gen_pool_free(uncached_pool[node], start, PAGE_SIZE);
 	}
-
 	return 0;
 }
 
 
-static int __init uncached_init(void) {
+static int __init uncached_init(void)
-	int i;
+{
+	int nid;
 
-	for (i = 0; i < MAX_NUMNODES; i++) {
+	for_each_online_node(nid) {
-		if (!node_online(i))
+		uncached_pool[nid] = gen_pool_create(PAGE_SHIFT, nid);
-			continue;
-		uncached_pool[i] = gen_pool_create(0, IA64_GRANULE_SHIFT,
-						   &uncached_get_new_chunk, i);
 	}
 
-	efi_memmap_walk_uc(uncached_build_memmap);
+	efi_memmap_walk_uc(uncached_build_memmap, NULL);
-
 	return 0;
 }
 

arch/ia64/sn/kernel/sn2/cache.c

@@ -3,11 +3,12 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  * 
- * Copyright (C) 2001-2003 Silicon Graphics, Inc. All rights reserved.
+ * Copyright (C) 2001-2003, 2006 Silicon Graphics, Inc. All rights reserved.
  *
  */
 #include <linux/module.h>
 #include <asm/pgalloc.h>
+#include <asm/sn/arch.h>
 
 /**
  * sn_flush_all_caches - flush a range of address from all caches (incl. L4)
@@ -17,18 +18,24 @@
  * Flush a range of addresses from all caches including L4. 
  * All addresses fully or partially contained within 
  * @flush_addr to @flush_addr + @bytes are flushed
- * from the all caches.
+ * from all caches.
  */
 void
 sn_flush_all_caches(long flush_addr, long bytes)
 {
-	flush_icache_range(flush_addr, flush_addr+bytes);
+	unsigned long addr = flush_addr;
+
+	/* SHub1 requires a cached address */
+	if (is_shub1() && (addr & RGN_BITS) == RGN_BASE(RGN_UNCACHED))
+		addr = (addr - RGN_BASE(RGN_UNCACHED)) + RGN_BASE(RGN_KERNEL);
+
+	flush_icache_range(addr, addr + bytes);
 	/*
 	 * The last call may have returned before the caches
 	 * were actually flushed, so we call it again to make
 	 * sure.
 	 */
-	flush_icache_range(flush_addr, flush_addr+bytes);
+	flush_icache_range(addr, addr + bytes);
 	mb();
 }
 EXPORT_SYMBOL(sn_flush_all_caches);

include/linux/genalloc.h

@@ -4,37 +4,32 @@
  * Uses for this includes on-device special memory, uncached memory
  * etc.
  *
- * This code is based on the buddy allocator found in the sym53c8xx_2
- * driver, adapted for general purpose use.
- *
  * This source code is licensed under the GNU General Public License,
  * Version 2.  See the file COPYING for more details.
  */
 
-#include <linux/spinlock.h>
-
-#define ALLOC_MIN_SHIFT		5 /* 32 bytes minimum */
-/*
- *  Link between free memory chunks of a given size.
- */
-struct gen_pool_link {
-	struct gen_pool_link *next;
-};
 
 /*
- *  Memory pool descriptor.
+ *  General purpose special memory pool descriptor.
  */
 struct gen_pool {
-	spinlock_t lock;
+	rwlock_t lock;
-	unsigned long (*get_new_chunk)(struct gen_pool *);
+	struct list_head chunks;	/* list of chunks in this pool */
-	struct gen_pool *next;
+	int min_alloc_order;		/* minimum allocation order */
-	struct gen_pool_link *h;
-	unsigned long private;
-	int max_chunk_shift;
 };
 
-unsigned long gen_pool_alloc(struct gen_pool *poolp, int size);
+/*
-void gen_pool_free(struct gen_pool *mp, unsigned long ptr, int size);
+ *  General purpose special memory pool chunk descriptor.
-struct gen_pool *gen_pool_create(int nr_chunks, int max_chunk_shift,
+ */
-				 unsigned long (*fp)(struct gen_pool *),
+struct gen_pool_chunk {
-				 unsigned long data);
+	spinlock_t lock;
+	struct list_head next_chunk;	/* next chunk in pool */
+	unsigned long start_addr;	/* starting address of memory chunk */
+	unsigned long end_addr;		/* ending address of memory chunk */
+	unsigned long bits[0];		/* bitmap for allocating memory chunk */
+};
+
+extern struct gen_pool *gen_pool_create(int, int);
+extern int gen_pool_add(struct gen_pool *, unsigned long, size_t, int);
+extern unsigned long gen_pool_alloc(struct gen_pool *, size_t);
+extern void gen_pool_free(struct gen_pool *, unsigned long, size_t);

lib/genalloc.c

@@ -4,10 +4,6 @@
  * Uses for this includes on-device special memory, uncached memory
  * etc.
  *
- * This code is based on the buddy allocator found in the sym53c8xx_2
- * driver Copyright (C) 1999-2001  Gerard Roudier <groudier@free.fr>,
- * and adapted for general purpose use.
- *
  * Copyright 2005 (C) Jes Sorensen <jes@trained-monkey.org>
  *
  * This source code is licensed under the GNU General Public License,
@@ -15,172 +11,155 @@
  */
 
 #include <linux/module.h>
-#include <linux/stddef.h>
-#include <linux/kernel.h>
-#include <linux/string.h>
-#include <linux/slab.h>
-#include <linux/init.h>
-#include <linux/mm.h>
-#include <linux/spinlock.h>
 #include <linux/genalloc.h>
 
-#include <asm/page.h>
-
-
-struct gen_pool *gen_pool_create(int nr_chunks, int max_chunk_shift,
-				 unsigned long (*fp)(struct gen_pool *),
-				 unsigned long data)
-{
-	struct gen_pool *poolp;
-	unsigned long tmp;
-	int i;
 
 /*
-	 * This is really an arbitrary limit, +10 is enough for
+ * Create a new special memory pool.
-	 * IA64_GRANULE_SHIFT, aka 16MB. If anyone needs a large limit
+ *
-	 * this can be increased without problems.
+ * @min_alloc_order: log base 2 of number of bytes each bitmap bit represents
+ * @nid: node id of the node the pool structure should be allocated on, or -1
  */
-	if ((max_chunk_shift > (PAGE_SHIFT + 10)) ||
+struct gen_pool *gen_pool_create(int min_alloc_order, int nid)
-	    ((max_chunk_shift < ALLOC_MIN_SHIFT) && max_chunk_shift))
+{
-		return NULL;
+	struct gen_pool *pool;
 
-	if (!max_chunk_shift)
+	pool = kmalloc_node(sizeof(struct gen_pool), GFP_KERNEL, nid);
-		max_chunk_shift = PAGE_SHIFT;
+	if (pool != NULL) {
-
+		rwlock_init(&pool->lock);
-	poolp = kmalloc(sizeof(struct gen_pool), GFP_KERNEL);
+		INIT_LIST_HEAD(&pool->chunks);
-	if (!poolp)
+		pool->min_alloc_order = min_alloc_order;
-		return NULL;
-	memset(poolp, 0, sizeof(struct gen_pool));
-	poolp->h = kmalloc(sizeof(struct gen_pool_link) *
-			   (max_chunk_shift - ALLOC_MIN_SHIFT + 1),
-			   GFP_KERNEL);
-	if (!poolp->h) {
-		printk(KERN_WARNING "gen_pool_alloc() failed to allocate\n");
-		kfree(poolp);
-		return NULL;
 	}
-	memset(poolp->h, 0, sizeof(struct gen_pool_link) *
+	return pool;
-	       (max_chunk_shift - ALLOC_MIN_SHIFT + 1));
-
-	spin_lock_init(&poolp->lock);
-	poolp->get_new_chunk = fp;
-	poolp->max_chunk_shift = max_chunk_shift;
-	poolp->private = data;
-
-	for (i = 0; i < nr_chunks; i++) {
-		tmp = poolp->get_new_chunk(poolp);
-		printk(KERN_INFO "allocated %lx\n", tmp);
-		if (!tmp)
-			break;
-		gen_pool_free(poolp, tmp, (1 << poolp->max_chunk_shift));
-	}
-
-	return poolp;
 }
 EXPORT_SYMBOL(gen_pool_create);
 
 
 /*
- *  Simple power of two buddy-like generic allocator.
+ * Add a new chunk of memory to the specified pool.
- *  Provides naturally aligned memory chunks.
+ *
+ * @pool: pool to add new memory chunk to
+ * @addr: starting address of memory chunk to add to pool
+ * @size: size in bytes of the memory chunk to add to pool
+ * @nid: node id of the node the chunk structure and bitmap should be
+ *       allocated on, or -1
  */
-unsigned long gen_pool_alloc(struct gen_pool *poolp, int size)
+int gen_pool_add(struct gen_pool *pool, unsigned long addr, size_t size,
+		 int nid)
 {
-	int j, i, s, max_chunk_size;
+	struct gen_pool_chunk *chunk;
-	unsigned long a, flags;
+	int nbits = size >> pool->min_alloc_order;
-	struct gen_pool_link *h = poolp->h;
+	int nbytes = sizeof(struct gen_pool_chunk) +
+				(nbits + BITS_PER_BYTE - 1) / BITS_PER_BYTE;
 
-	max_chunk_size = 1 << poolp->max_chunk_shift;
+	chunk = kmalloc_node(nbytes, GFP_KERNEL, nid);
+	if (unlikely(chunk == NULL))
+		return -1;
 
-	if (size > max_chunk_size)
+	memset(chunk, 0, nbytes);
+	spin_lock_init(&chunk->lock);
+	chunk->start_addr = addr;
+	chunk->end_addr = addr + size;
+
+	write_lock(&pool->lock);
+	list_add(&chunk->next_chunk, &pool->chunks);
+	write_unlock(&pool->lock);
+
+	return 0;
+}
+EXPORT_SYMBOL(gen_pool_add);
+
+
+/*
+ * Allocate the requested number of bytes from the specified pool.
+ * Uses a first-fit algorithm.
+ *
+ * @pool: pool to allocate from
+ * @size: number of bytes to allocate from the pool
+ */
+unsigned long gen_pool_alloc(struct gen_pool *pool, size_t size)
+{
+	struct list_head *_chunk;
+	struct gen_pool_chunk *chunk;
+	unsigned long addr, flags;
+	int order = pool->min_alloc_order;
+	int nbits, bit, start_bit, end_bit;
+
+	if (size == 0)
 		return 0;
 
-	size = max(size, 1 << ALLOC_MIN_SHIFT);
+	nbits = (size + (1UL << order) - 1) >> order;
-	i = fls(size - 1);
-	s = 1 << i;
-	j = i -= ALLOC_MIN_SHIFT;
 
-	spin_lock_irqsave(&poolp->lock, flags);
+	read_lock(&pool->lock);
-	while (!h[j].next) {
+	list_for_each(_chunk, &pool->chunks) {
-		if (s == max_chunk_size) {
+		chunk = list_entry(_chunk, struct gen_pool_chunk, next_chunk);
-			struct gen_pool_link *ptr;
+
-			spin_unlock_irqrestore(&poolp->lock, flags);
+		end_bit = (chunk->end_addr - chunk->start_addr) >> order;
-			ptr = (struct gen_pool_link *)poolp->get_new_chunk(poolp);
+		end_bit -= nbits + 1;
-			spin_lock_irqsave(&poolp->lock, flags);
+
-			h[j].next = ptr;
+		spin_lock_irqsave(&chunk->lock, flags);
-			if (h[j].next)
+		bit = -1;
-				h[j].next->next = NULL;
+		while (bit + 1 < end_bit) {
+			bit = find_next_zero_bit(chunk->bits, end_bit, bit + 1);
+			if (bit >= end_bit)
 				break;
+
+			start_bit = bit;
+			if (nbits > 1) {
+				bit = find_next_bit(chunk->bits, bit + nbits,
+							bit + 1);
+				if (bit - start_bit < nbits)
+					continue;
 			}
-		j++;
+
-		s <<= 1;
+			addr = chunk->start_addr +
+					    ((unsigned long)start_bit << order);
+			while (nbits--)
+				__set_bit(start_bit++, &chunk->bits);
+			spin_unlock_irqrestore(&chunk->lock, flags);
+			read_unlock(&pool->lock);
+			return addr;
 		}
-	a = (unsigned long) h[j].next;
+		spin_unlock_irqrestore(&chunk->lock, flags);
-	if (a) {
-		h[j].next = h[j].next->next;
-		/*
-		 * This should be split into a seperate function doing
-		 * the chunk split in order to support custom
-		 * handling memory not physically accessible by host
-		 */
-		while (j > i) {
-			j -= 1;
-			s >>= 1;
-			h[j].next = (struct gen_pool_link *) (a + s);
-			h[j].next->next = NULL;
 	}
-	}
+	read_unlock(&pool->lock);
-	spin_unlock_irqrestore(&poolp->lock, flags);
+	return 0;
-	return a;
 }
 EXPORT_SYMBOL(gen_pool_alloc);
 
 
 /*
- *  Counter-part of the generic allocator.
+ * Free the specified memory back to the specified pool.
+ *
+ * @pool: pool to free to
+ * @addr: starting address of memory to free back to pool
+ * @size: size in bytes of memory to free
  */
-void gen_pool_free(struct gen_pool *poolp, unsigned long ptr, int size)
+void gen_pool_free(struct gen_pool *pool, unsigned long addr, size_t size)
 {
-	struct gen_pool_link *q;
+	struct list_head *_chunk;
-	struct gen_pool_link *h = poolp->h;
+	struct gen_pool_chunk *chunk;
-	unsigned long a, b, flags;
+	unsigned long flags;
-	int i, s, max_chunk_size;
+	int order = pool->min_alloc_order;
+	int bit, nbits;
 
-	max_chunk_size = 1 << poolp->max_chunk_shift;
+	nbits = (size + (1UL << order) - 1) >> order;
 
-	if (size > max_chunk_size)
+	read_lock(&pool->lock);
-		return;
+	list_for_each(_chunk, &pool->chunks) {
+		chunk = list_entry(_chunk, struct gen_pool_chunk, next_chunk);
 
-	size = max(size, 1 << ALLOC_MIN_SHIFT);
+		if (addr >= chunk->start_addr && addr < chunk->end_addr) {
-	i = fls(size - 1);
+			BUG_ON(addr + size > chunk->end_addr);
-	s = 1 << i;
+			spin_lock_irqsave(&chunk->lock, flags);
-	i -= ALLOC_MIN_SHIFT;
+			bit = (addr - chunk->start_addr) >> order;
-
+			while (nbits--)
-	a = ptr;
+				__clear_bit(bit++, &chunk->bits);
-
+			spin_unlock_irqrestore(&chunk->lock, flags);
-	spin_lock_irqsave(&poolp->lock, flags);
-	while (1) {
-		if (s == max_chunk_size) {
-			((struct gen_pool_link *)a)->next = h[i].next;
-			h[i].next = (struct gen_pool_link *)a;
 			break;
 		}
-		b = a ^ s;
-		q = &h[i];
-
-		while (q->next && q->next != (struct gen_pool_link *)b)
-			q = q->next;
-
-		if (!q->next) {
-			((struct gen_pool_link *)a)->next = h[i].next;
-			h[i].next = (struct gen_pool_link *)a;
-			break;
 	}
-		q->next = q->next->next;
+	BUG_ON(nbits > 0);
-		a = a & b;
+	read_unlock(&pool->lock);
-		s <<= 1;
-		i++;
-	}
-	spin_unlock_irqrestore(&poolp->lock, flags);
 }
 EXPORT_SYMBOL(gen_pool_free);