X-Git-Url: http://pilppa.org/gitweb/gitweb.cgi?a=blobdiff_plain;f=fs%2Fext4%2Fmballoc.c;h=4865d318d7e1631c871e0985dfe66a427976275c;hb=e8546d0615542684ca02ba03edebec1a503beb6b;hp=76e5fedc0a0b4c7e581852d9eeab4374ba246a24;hpb=c1f3ee120bb61045b1c0a3ead620d1d65af47130;p=linux-2.6-omap-h63xx.git

diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 76e5fedc0a0..4865d318d7e 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -420,6 +420,7 @@
 #define MB_DEFAULT_GROUP_PREALLOC	512
 
 static struct kmem_cache *ext4_pspace_cachep;
+static struct kmem_cache *ext4_ac_cachep;
 
 #ifdef EXT4_BB_MAX_BLOCKS
 #undef EXT4_BB_MAX_BLOCKS
@@ -626,21 +627,19 @@ static ext4_fsblk_t ext4_grp_offs_to_block(struct super_block *sb,
 	return block;
 }
 
+static inline void *mb_correct_addr_and_bit(int *bit, void *addr)
+{
 #if BITS_PER_LONG == 64
-#define mb_correct_addr_and_bit(bit, addr)		\
-{							\
-	bit += ((unsigned long) addr & 7UL) << 3;	\
-	addr = (void *) ((unsigned long) addr & ~7UL);	\
-}
+	*bit += ((unsigned long) addr & 7UL) << 3;
+	addr = (void *) ((unsigned long) addr & ~7UL);
 #elif BITS_PER_LONG == 32
-#define mb_correct_addr_and_bit(bit, addr)		\
-{							\
-	bit += ((unsigned long) addr & 3UL) << 3;	\
-	addr = (void *) ((unsigned long) addr & ~3UL);	\
-}
+	*bit += ((unsigned long) addr & 3UL) << 3;
+	addr = (void *) ((unsigned long) addr & ~3UL);
 #else
 #error "how many bits you are?!"
 #endif
+	return addr;
+}
 
 static inline int mb_test_bit(int bit, void *addr)
 {
@@ -648,39 +647,58 @@ static inline int mb_test_bit(int bit, void *addr)
 	 * ext4_test_bit on architecture like powerpc
 	 * needs unsigned long aligned address
 	 */
-	mb_correct_addr_and_bit(bit, addr);
+	addr = mb_correct_addr_and_bit(&bit, addr);
 	return ext4_test_bit(bit, addr);
 }
 
 static inline void mb_set_bit(int bit, void *addr)
 {
-	mb_correct_addr_and_bit(bit, addr);
+	addr = mb_correct_addr_and_bit(&bit, addr);
 	ext4_set_bit(bit, addr);
 }
 
 static inline void mb_set_bit_atomic(spinlock_t *lock, int bit, void *addr)
 {
-	mb_correct_addr_and_bit(bit, addr);
+	addr = mb_correct_addr_and_bit(&bit, addr);
 	ext4_set_bit_atomic(lock, bit, addr);
 }
 
 static inline void mb_clear_bit(int bit, void *addr)
 {
-	mb_correct_addr_and_bit(bit, addr);
+	addr = mb_correct_addr_and_bit(&bit, addr);
 	ext4_clear_bit(bit, addr);
 }
 
 static inline void mb_clear_bit_atomic(spinlock_t *lock, int bit, void *addr)
 {
-	mb_correct_addr_and_bit(bit, addr);
+	addr = mb_correct_addr_and_bit(&bit, addr);
 	ext4_clear_bit_atomic(lock, bit, addr);
 }
 
+static inline int mb_find_next_zero_bit(void *addr, int max, int start)
+{
+	int fix = 0;
+	addr = mb_correct_addr_and_bit(&fix, addr);
+	max += fix;
+	start += fix;
+
+	return ext4_find_next_zero_bit(addr, max, start) - fix;
+}
+
+static inline int mb_find_next_bit(void *addr, int max, int start)
+{
+	int fix = 0;
+	addr = mb_correct_addr_and_bit(&fix, addr);
+	max += fix;
+	start += fix;
+
+	return ext4_find_next_bit(addr, max, start) - fix;
+}
+
 static void *mb_find_buddy(struct ext4_buddy *e4b, int order, int *max)
 {
 	char *bb;
 
-	/* FIXME!! is this needed */
 	BUG_ON(EXT4_MB_BITMAP(e4b) == EXT4_MB_BUDDY(e4b));
 	BUG_ON(max == NULL);
 
@@ -906,7 +924,7 @@ static void ext4_mb_mark_free_simple(struct super_block *sb,
 	unsigned short chunk;
 	unsigned short border;
 
-	BUG_ON(len >= EXT4_BLOCKS_PER_GROUP(sb));
+	BUG_ON(len > EXT4_BLOCKS_PER_GROUP(sb));
 
 	border = 2 << sb->s_blocksize_bits;
 
@@ -946,12 +964,12 @@ static void ext4_mb_generate_buddy(struct super_block *sb,
 
 	/* initialize buddy from bitmap which is aggregation
 	 * of on-disk bitmap and preallocations */
-	i = ext4_find_next_zero_bit(bitmap, max, 0);
+	i = mb_find_next_zero_bit(bitmap, max, 0);
 	grp->bb_first_free = i;
 	while (i < max) {
 		fragments++;
 		first = i;
-		i = ext4_find_next_bit(bitmap, max, i);
+		i = mb_find_next_bit(bitmap, max, i);
 		len = i - first;
 		free += len;
 		if (len > 1)
@@ -959,14 +977,18 @@ static void ext4_mb_generate_buddy(struct super_block *sb,
 		else
 			grp->bb_counters[0]++;
 		if (i < max)
-			i = ext4_find_next_zero_bit(bitmap, max, i);
+			i = mb_find_next_zero_bit(bitmap, max, i);
 	}
 	grp->bb_fragments = fragments;
 
 	if (free != grp->bb_free) {
-		printk(KERN_DEBUG
+		ext4_error(sb, __FUNCTION__,
 			"EXT4-fs: group %lu: %u blocks in bitmap, %u in gd\n",
 			group, free, grp->bb_free);
+		/*
+		 * If we intent to continue, we consider group descritor
+		 * corrupt and update bb_free using bitmap value
+		 */
 		grp->bb_free = free;
 	}
 
@@ -1146,8 +1168,9 @@ out:
 	return err;
 }
 
-static int ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
-		struct ext4_buddy *e4b)
+static noinline_for_stack int
+ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
+					struct ext4_buddy *e4b)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	struct inode *inode = sbi->s_buddy_cache;
@@ -1778,7 +1801,7 @@ static void ext4_mb_simple_scan_group(struct ext4_allocation_context *ac,
 		buddy = mb_find_buddy(e4b, i, &max);
 		BUG_ON(buddy == NULL);
 
-		k = ext4_find_next_zero_bit(buddy, max, 0);
+		k = mb_find_next_zero_bit(buddy, max, 0);
 		BUG_ON(k >= max);
 
 		ac->ac_found++;
@@ -1818,16 +1841,33 @@ static void ext4_mb_complex_scan_group(struct ext4_allocation_context *ac,
 	i = e4b->bd_info->bb_first_free;
 
 	while (free && ac->ac_status == AC_STATUS_CONTINUE) {
-		i = ext4_find_next_zero_bit(bitmap,
+		i = mb_find_next_zero_bit(bitmap,
 						EXT4_BLOCKS_PER_GROUP(sb), i);
 		if (i >= EXT4_BLOCKS_PER_GROUP(sb)) {
-			BUG_ON(free != 0);
+			/*
+			 * IF we have corrupt bitmap, we won't find any
+			 * free blocks even though group info says we
+			 * we have free blocks
+			 */
+			ext4_error(sb, __FUNCTION__, "%d free blocks as per "
+					"group info. But bitmap says 0\n",
+					free);
 			break;
 		}
 
 		mb_find_extent(e4b, 0, i, ac->ac_g_ex.fe_len, &ex);
 		BUG_ON(ex.fe_len <= 0);
-		BUG_ON(free < ex.fe_len);
+		if (free < ex.fe_len) {
+			ext4_error(sb, __FUNCTION__, "%d free blocks as per "
+					"group info. But got %d blocks\n",
+					free, ex.fe_len);
+			/*
+			 * The number of free blocks differs. This mostly
+			 * indicate that the bitmap is corrupt. So exit
+			 * without claiming the space.
+			 */
+			break;
+		}
 
 		ext4_mb_measure_extent(ac, &ex, e4b);
 
@@ -1926,7 +1966,8 @@ static int ext4_mb_good_group(struct ext4_allocation_context *ac,
 	return 0;
 }
 
-static int ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
+static noinline_for_stack int
+ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 {
 	ext4_group_t group;
 	ext4_group_t i;
@@ -2410,17 +2451,10 @@ static void ext4_mb_history_init(struct super_block *sb)
 	int i;
 
 	if (sbi->s_mb_proc != NULL) {
-		struct proc_dir_entry *p;
-		p = create_proc_entry("mb_history", S_IRUGO, sbi->s_mb_proc);
-		if (p) {
-			p->proc_fops = &ext4_mb_seq_history_fops;
-			p->data = sb;
-		}
-		p = create_proc_entry("mb_groups", S_IRUGO, sbi->s_mb_proc);
-		if (p) {
-			p->proc_fops = &ext4_mb_seq_groups_fops;
-			p->data = sb;
-		}
+		proc_create_data("mb_history", S_IRUGO, sbi->s_mb_proc,
+				 &ext4_mb_seq_history_fops, sb);
+		proc_create_data("mb_groups", S_IRUGO, sbi->s_mb_proc,
+				 &ext4_mb_seq_groups_fops, sb);
 	}
 
 	sbi->s_mb_history_max = 1000;
@@ -2433,7 +2467,8 @@ static void ext4_mb_history_init(struct super_block *sb)
 	/* if we can't allocate history, then we simple won't use it */
 }
 
-static void ext4_mb_store_history(struct ext4_allocation_context *ac)
+static noinline_for_stack void
+ext4_mb_store_history(struct ext4_allocation_context *ac)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(ac->ac_sb);
 	struct ext4_mb_history h;
@@ -2769,7 +2804,8 @@ int ext4_mb_release(struct super_block *sb)
 	return 0;
 }
 
-static void ext4_mb_free_committed_blocks(struct super_block *sb)
+static noinline_for_stack void
+ext4_mb_free_committed_blocks(struct super_block *sb)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	int err;
@@ -2828,7 +2864,6 @@ static void ext4_mb_free_committed_blocks(struct super_block *sb)
 	mb_debug("freed %u blocks in %u structures\n", count, count2);
 }
 
-#define EXT4_ROOT			"ext4"
 #define EXT4_MB_STATS_NAME		"stats"
 #define EXT4_MB_MAX_TO_SCAN_NAME	"max_to_scan"
 #define EXT4_MB_MIN_TO_SCAN_NAME	"min_to_scan"
@@ -2959,12 +2994,19 @@ int __init init_ext4_mballoc(void)
 	if (ext4_pspace_cachep == NULL)
 		return -ENOMEM;
 
+	ext4_ac_cachep =
+		kmem_cache_create("ext4_alloc_context",
+				     sizeof(struct ext4_allocation_context),
+				     0, SLAB_RECLAIM_ACCOUNT, NULL);
+	if (ext4_ac_cachep == NULL) {
+		kmem_cache_destroy(ext4_pspace_cachep);
+		return -ENOMEM;
+	}
 #ifdef CONFIG_PROC_FS
-	proc_root_ext4 = proc_mkdir(EXT4_ROOT, proc_root_fs);
+	proc_root_ext4 = proc_mkdir("fs/ext4", NULL);
 	if (proc_root_ext4 == NULL)
-		printk(KERN_ERR "EXT4-fs: Unable to create %s\n", EXT4_ROOT);
+		printk(KERN_ERR "EXT4-fs: Unable to create fs/ext4\n");
 #endif
-
 	return 0;
 }
 
@@ -2972,8 +3014,9 @@ void exit_ext4_mballoc(void)
 {
 	/* XXX: synchronize_rcu(); */
 	kmem_cache_destroy(ext4_pspace_cachep);
+	kmem_cache_destroy(ext4_ac_cachep);
 #ifdef CONFIG_PROC_FS
-	remove_proc_entry(EXT4_ROOT, proc_root_fs);
+	remove_proc_entry("fs/ext4", NULL);
 #endif
 }
 
@@ -2982,7 +3025,8 @@ void exit_ext4_mballoc(void)
  * Check quota and mark choosed space (ac->ac_b_ex) non-free in bitmaps
  * Returns 0 if success or error code
  */
-static int ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
+static noinline_for_stack int
+ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
 				handle_t *handle)
 {
 	struct buffer_head *bitmap_bh = NULL;
@@ -3055,9 +3099,7 @@ static int ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
 						ac->ac_b_ex.fe_group,
 						gdp));
 	}
-	gdp->bg_free_blocks_count =
-		cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count)
-				- ac->ac_b_ex.fe_len);
+	le16_add_cpu(&gdp->bg_free_blocks_count, -ac->ac_b_ex.fe_len);
 	gdp->bg_checksum = ext4_group_desc_csum(sbi, ac->ac_b_ex.fe_group, gdp);
 	spin_unlock(sb_bgl_lock(sbi, ac->ac_b_ex.fe_group));
 	percpu_counter_sub(&sbi->s_freeblocks_counter, ac->ac_b_ex.fe_len);
@@ -3069,7 +3111,7 @@ static int ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
 
 out_err:
 	sb->s_dirt = 1;
-	put_bh(bitmap_bh);
+	brelse(bitmap_bh);
 	return err;
 }
 
@@ -3099,15 +3141,16 @@ static void ext4_mb_normalize_group_request(struct ext4_allocation_context *ac)
  * Normalization means making request better in terms of
  * size and alignment
  */
-static void ext4_mb_normalize_request(struct ext4_allocation_context *ac,
+static noinline_for_stack void
+ext4_mb_normalize_request(struct ext4_allocation_context *ac,
 				struct ext4_allocation_request *ar)
 {
 	int bsbits, max;
 	ext4_lblk_t end;
-	struct list_head *cur;
 	loff_t size, orig_size, start_off;
 	ext4_lblk_t start, orig_start;
 	struct ext4_inode_info *ei = EXT4_I(ac->ac_inode);
+	struct ext4_prealloc_space *pa;
 
 	/* do normalize only data requests, metadata requests
 	   do not need preallocation */
@@ -3193,12 +3236,9 @@ static void ext4_mb_normalize_request(struct ext4_allocation_context *ac,
 
 	/* check we don't cross already preallocated blocks */
 	rcu_read_lock();
-	list_for_each_rcu(cur, &ei->i_prealloc_list) {
-		struct ext4_prealloc_space *pa;
+	list_for_each_entry_rcu(pa, &ei->i_prealloc_list, pa_inode_list) {
 		unsigned long pa_end;
 
-		pa = list_entry(cur, struct ext4_prealloc_space, pa_inode_list);
-
 		if (pa->pa_deleted)
 			continue;
 		spin_lock(&pa->pa_lock);
@@ -3240,10 +3280,8 @@ static void ext4_mb_normalize_request(struct ext4_allocation_context *ac,
 
 	/* XXX: extra loop to check we really don't overlap preallocations */
 	rcu_read_lock();
-	list_for_each_rcu(cur, &ei->i_prealloc_list) {
-		struct ext4_prealloc_space *pa;
+	list_for_each_entry_rcu(pa, &ei->i_prealloc_list, pa_inode_list) {
 		unsigned long pa_end;
-		pa = list_entry(cur, struct ext4_prealloc_space, pa_inode_list);
 		spin_lock(&pa->pa_lock);
 		if (pa->pa_deleted == 0) {
 			pa_end = pa->pa_lstart + pa->pa_len;
@@ -3354,13 +3392,10 @@ static void ext4_mb_use_group_pa(struct ext4_allocation_context *ac,
 	ac->ac_pa = pa;
 
 	/* we don't correct pa_pstart or pa_plen here to avoid
-	 * possible race when tte group is being loaded concurrently
+	 * possible race when the group is being loaded concurrently
 	 * instead we correct pa later, after blocks are marked
-	 * in on-disk bitmap -- see ext4_mb_release_context() */
-	/*
-	 * FIXME!! but the other CPUs can look at this particular
-	 * pa and think that it have enought free blocks if we
-	 * don't update pa_free here right ?
+	 * in on-disk bitmap -- see ext4_mb_release_context()
+	 * Other CPUs are prevented from allocating from this pa by lg_mutex
 	 */
 	mb_debug("use %u/%u from group pa %p\n", pa->pa_lstart-len, len, pa);
 }
@@ -3368,12 +3403,12 @@ static void ext4_mb_use_group_pa(struct ext4_allocation_context *ac,
 /*
  * search goal blocks in preallocated space
  */
-static int ext4_mb_use_preallocated(struct ext4_allocation_context *ac)
+static noinline_for_stack int
+ext4_mb_use_preallocated(struct ext4_allocation_context *ac)
 {
 	struct ext4_inode_info *ei = EXT4_I(ac->ac_inode);
 	struct ext4_locality_group *lg;
 	struct ext4_prealloc_space *pa;
-	struct list_head *cur;
 
 	/* only data can be preallocated */
 	if (!(ac->ac_flags & EXT4_MB_HINT_DATA))
@@ -3381,8 +3416,7 @@ static int ext4_mb_use_preallocated(struct ext4_allocation_context *ac)
 
 	/* first, try per-file preallocation */
 	rcu_read_lock();
-	list_for_each_rcu(cur, &ei->i_prealloc_list) {
-		pa = list_entry(cur, struct ext4_prealloc_space, pa_inode_list);
+	list_for_each_entry_rcu(pa, &ei->i_prealloc_list, pa_inode_list) {
 
 		/* all fields in this condition don't change,
 		 * so we can skip locking for them */
@@ -3414,8 +3448,7 @@ static int ext4_mb_use_preallocated(struct ext4_allocation_context *ac)
 		return 0;
 
 	rcu_read_lock();
-	list_for_each_rcu(cur, &lg->lg_prealloc_list) {
-		pa = list_entry(cur, struct ext4_prealloc_space, pa_inode_list);
+	list_for_each_entry_rcu(pa, &lg->lg_prealloc_list, pa_inode_list) {
 		spin_lock(&pa->pa_lock);
 		if (pa->pa_deleted == 0 && pa->pa_free >= ac->ac_o_ex.fe_len) {
 			atomic_inc(&pa->pa_count);
@@ -3535,7 +3568,8 @@ static void ext4_mb_put_pa(struct ext4_allocation_context *ac,
 /*
  * creates new preallocated space for given inode
  */
-static int ext4_mb_new_inode_pa(struct ext4_allocation_context *ac)
+static noinline_for_stack int
+ext4_mb_new_inode_pa(struct ext4_allocation_context *ac)
 {
 	struct super_block *sb = ac->ac_sb;
 	struct ext4_prealloc_space *pa;
@@ -3622,7 +3656,8 @@ static int ext4_mb_new_inode_pa(struct ext4_allocation_context *ac)
 /*
  * creates new preallocated space for locality group inodes belongs to
  */
-static int ext4_mb_new_group_pa(struct ext4_allocation_context *ac)
+static noinline_for_stack int
+ext4_mb_new_group_pa(struct ext4_allocation_context *ac)
 {
 	struct super_block *sb = ac->ac_sb;
 	struct ext4_locality_group *lg;
@@ -3695,11 +3730,11 @@ static int ext4_mb_new_preallocation(struct ext4_allocation_context *ac)
  * the caller MUST hold group/inode locks.
  * TODO: optimize the case when there are no in-core structures yet
  */
-static int ext4_mb_release_inode_pa(struct ext4_buddy *e4b,
-				struct buffer_head *bitmap_bh,
+static noinline_for_stack int
+ext4_mb_release_inode_pa(struct ext4_buddy *e4b, struct buffer_head *bitmap_bh,
 				struct ext4_prealloc_space *pa)
 {
-	struct ext4_allocation_context ac;
+	struct ext4_allocation_context *ac;
 	struct super_block *sb = e4b->bd_sb;
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	unsigned long end;
@@ -3715,15 +3750,19 @@ static int ext4_mb_release_inode_pa(struct ext4_buddy *e4b,
 	BUG_ON(group != e4b->bd_group && pa->pa_len != 0);
 	end = bit + pa->pa_len;
 
-	ac.ac_sb = sb;
-	ac.ac_inode = pa->pa_inode;
-	ac.ac_op = EXT4_MB_HISTORY_DISCARD;
+	ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
+
+	if (ac) {
+		ac->ac_sb = sb;
+		ac->ac_inode = pa->pa_inode;
+		ac->ac_op = EXT4_MB_HISTORY_DISCARD;
+	}
 
 	while (bit < end) {
-		bit = ext4_find_next_zero_bit(bitmap_bh->b_data, end, bit);
+		bit = mb_find_next_zero_bit(bitmap_bh->b_data, end, bit);
 		if (bit >= end)
 			break;
-		next = ext4_find_next_bit(bitmap_bh->b_data, end, bit);
+		next = mb_find_next_bit(bitmap_bh->b_data, end, bit);
 		if (next > end)
 			next = end;
 		start = group * EXT4_BLOCKS_PER_GROUP(sb) + bit +
@@ -3733,37 +3772,49 @@ static int ext4_mb_release_inode_pa(struct ext4_buddy *e4b,
 				(unsigned) group);
 		free += next - bit;
 
-		ac.ac_b_ex.fe_group = group;
-		ac.ac_b_ex.fe_start = bit;
-		ac.ac_b_ex.fe_len = next - bit;
-		ac.ac_b_ex.fe_logical = 0;
-		ext4_mb_store_history(&ac);
+		if (ac) {
+			ac->ac_b_ex.fe_group = group;
+			ac->ac_b_ex.fe_start = bit;
+			ac->ac_b_ex.fe_len = next - bit;
+			ac->ac_b_ex.fe_logical = 0;
+			ext4_mb_store_history(ac);
+		}
 
 		mb_free_blocks(pa->pa_inode, e4b, bit, next - bit);
 		bit = next + 1;
 	}
 	if (free != pa->pa_free) {
-		printk(KERN_ERR "pa %p: logic %lu, phys. %lu, len %lu\n",
+		printk(KERN_CRIT "pa %p: logic %lu, phys. %lu, len %lu\n",
 			pa, (unsigned long) pa->pa_lstart,
 			(unsigned long) pa->pa_pstart,
 			(unsigned long) pa->pa_len);
-		printk(KERN_ERR "free %u, pa_free %u\n", free, pa->pa_free);
+		ext4_error(sb, __FUNCTION__, "free %u, pa_free %u\n",
+						free, pa->pa_free);
+		/*
+		 * pa is already deleted so we use the value obtained
+		 * from the bitmap and continue.
+		 */
 	}
-	BUG_ON(free != pa->pa_free);
 	atomic_add(free, &sbi->s_mb_discarded);
+	if (ac)
+		kmem_cache_free(ext4_ac_cachep, ac);
 
 	return err;
 }
 
-static int ext4_mb_release_group_pa(struct ext4_buddy *e4b,
+static noinline_for_stack int
+ext4_mb_release_group_pa(struct ext4_buddy *e4b,
 				struct ext4_prealloc_space *pa)
 {
-	struct ext4_allocation_context ac;
+	struct ext4_allocation_context *ac;
 	struct super_block *sb = e4b->bd_sb;
 	ext4_group_t group;
 	ext4_grpblk_t bit;
 
-	ac.ac_op = EXT4_MB_HISTORY_DISCARD;
+	ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
+
+	if (ac)
+		ac->ac_op = EXT4_MB_HISTORY_DISCARD;
 
 	BUG_ON(pa->pa_deleted == 0);
 	ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, &bit);
@@ -3771,13 +3822,16 @@ static int ext4_mb_release_group_pa(struct ext4_buddy *e4b,
 	mb_free_blocks(pa->pa_inode, e4b, bit, pa->pa_len);
 	atomic_add(pa->pa_len, &EXT4_SB(sb)->s_mb_discarded);
 
-	ac.ac_sb = sb;
-	ac.ac_inode = NULL;
-	ac.ac_b_ex.fe_group = group;
-	ac.ac_b_ex.fe_start = bit;
-	ac.ac_b_ex.fe_len = pa->pa_len;
-	ac.ac_b_ex.fe_logical = 0;
-	ext4_mb_store_history(&ac);
+	if (ac) {
+		ac->ac_sb = sb;
+		ac->ac_inode = NULL;
+		ac->ac_b_ex.fe_group = group;
+		ac->ac_b_ex.fe_start = bit;
+		ac->ac_b_ex.fe_len = pa->pa_len;
+		ac->ac_b_ex.fe_logical = 0;
+		ext4_mb_store_history(ac);
+		kmem_cache_free(ext4_ac_cachep, ac);
+	}
 
 	return 0;
 }
@@ -3791,7 +3845,8 @@ static int ext4_mb_release_group_pa(struct ext4_buddy *e4b,
  * - how many do we discard
  *   1) how many requested
  */
-static int ext4_mb_discard_group_preallocations(struct super_block *sb,
+static noinline_for_stack int
+ext4_mb_discard_group_preallocations(struct super_block *sb,
 					ext4_group_t group, int needed)
 {
 	struct ext4_group_info *grp = ext4_get_group_info(sb, group);
@@ -4113,7 +4168,8 @@ static void ext4_mb_group_or_file(struct ext4_allocation_context *ac)
 	mutex_lock(&ac->ac_lg->lg_mutex);
 }
 
-static int ext4_mb_initialize_context(struct ext4_allocation_context *ac,
+static noinline_for_stack int
+ext4_mb_initialize_context(struct ext4_allocation_context *ac,
 				struct ext4_allocation_request *ar)
 {
 	struct super_block *sb = ar->inode->i_sb;
@@ -4231,7 +4287,7 @@ static int ext4_mb_discard_preallocations(struct super_block *sb, int needed)
 ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
 				 struct ext4_allocation_request *ar, int *errp)
 {
-	struct ext4_allocation_context ac;
+	struct ext4_allocation_context *ac = NULL;
 	struct ext4_sb_info *sbi;
 	struct super_block *sb;
 	ext4_fsblk_t block = 0;
@@ -4257,53 +4313,60 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
 	}
 	inquota = ar->len;
 
+	ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
+	if (!ac) {
+		*errp = -ENOMEM;
+		return 0;
+	}
+
 	ext4_mb_poll_new_transaction(sb, handle);
 
-	*errp = ext4_mb_initialize_context(&ac, ar);
+	*errp = ext4_mb_initialize_context(ac, ar);
 	if (*errp) {
 		ar->len = 0;
 		goto out;
 	}
 
-	ac.ac_op = EXT4_MB_HISTORY_PREALLOC;
-	if (!ext4_mb_use_preallocated(&ac)) {
+	ac->ac_op = EXT4_MB_HISTORY_PREALLOC;
+	if (!ext4_mb_use_preallocated(ac)) {
 
-		ac.ac_op = EXT4_MB_HISTORY_ALLOC;
-		ext4_mb_normalize_request(&ac, ar);
+		ac->ac_op = EXT4_MB_HISTORY_ALLOC;
+		ext4_mb_normalize_request(ac, ar);
 
 repeat:
 		/* allocate space in core */
-		ext4_mb_regular_allocator(&ac);
+		ext4_mb_regular_allocator(ac);
 
 		/* as we've just preallocated more space than
 		 * user requested orinally, we store allocated
 		 * space in a special descriptor */
-		if (ac.ac_status == AC_STATUS_FOUND &&
-				ac.ac_o_ex.fe_len < ac.ac_b_ex.fe_len)
-			ext4_mb_new_preallocation(&ac);
+		if (ac->ac_status == AC_STATUS_FOUND &&
+				ac->ac_o_ex.fe_len < ac->ac_b_ex.fe_len)
+			ext4_mb_new_preallocation(ac);
 	}
 
-	if (likely(ac.ac_status == AC_STATUS_FOUND)) {
-		ext4_mb_mark_diskspace_used(&ac, handle);
+	if (likely(ac->ac_status == AC_STATUS_FOUND)) {
+		ext4_mb_mark_diskspace_used(ac, handle);
 		*errp = 0;
-		block = ext4_grp_offs_to_block(sb, &ac.ac_b_ex);
-		ar->len = ac.ac_b_ex.fe_len;
+		block = ext4_grp_offs_to_block(sb, &ac->ac_b_ex);
+		ar->len = ac->ac_b_ex.fe_len;
 	} else {
-		freed  = ext4_mb_discard_preallocations(sb, ac.ac_o_ex.fe_len);
+		freed  = ext4_mb_discard_preallocations(sb, ac->ac_o_ex.fe_len);
 		if (freed)
 			goto repeat;
 		*errp = -ENOSPC;
-		ac.ac_b_ex.fe_len = 0;
+		ac->ac_b_ex.fe_len = 0;
 		ar->len = 0;
-		ext4_mb_show_ac(&ac);
+		ext4_mb_show_ac(ac);
 	}
 
-	ext4_mb_release_context(&ac);
+	ext4_mb_release_context(ac);
 
 out:
 	if (ar->len < inquota)
 		DQUOT_FREE_BLOCK(ar->inode, inquota - ar->len);
 
+	kmem_cache_free(ext4_ac_cachep, ac);
 	return block;
 }
 static void ext4_mb_poll_new_transaction(struct super_block *sb,
@@ -4337,7 +4400,8 @@ static void ext4_mb_poll_new_transaction(struct super_block *sb,
 	ext4_mb_free_committed_blocks(sb);
 }
 
-static int ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b,
+static noinline_for_stack int
+ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b,
 			  ext4_group_t group, ext4_grpblk_t block, int count)
 {
 	struct ext4_group_info *db = e4b->bd_info;
@@ -4405,9 +4469,9 @@ void ext4_mb_free_blocks(handle_t *handle, struct inode *inode,
 			unsigned long block, unsigned long count,
 			int metadata, unsigned long *freed)
 {
-	struct buffer_head *bitmap_bh = 0;
+	struct buffer_head *bitmap_bh = NULL;
 	struct super_block *sb = inode->i_sb;
-	struct ext4_allocation_context ac;
+	struct ext4_allocation_context *ac = NULL;
 	struct ext4_group_desc *gdp;
 	struct ext4_super_block *es;
 	unsigned long overflow;
@@ -4436,9 +4500,12 @@ void ext4_mb_free_blocks(handle_t *handle, struct inode *inode,
 
 	ext4_debug("freeing block %lu\n", block);
 
-	ac.ac_op = EXT4_MB_HISTORY_FREE;
-	ac.ac_inode = inode;
-	ac.ac_sb = sb;
+	ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
+	if (ac) {
+		ac->ac_op = EXT4_MB_HISTORY_FREE;
+		ac->ac_inode = inode;
+		ac->ac_sb = sb;
+	}
 
 do_more:
 	overflow = 0;
@@ -4504,10 +4571,12 @@ do_more:
 	BUFFER_TRACE(bitmap_bh, "dirtied bitmap block");
 	err = ext4_journal_dirty_metadata(handle, bitmap_bh);
 
-	ac.ac_b_ex.fe_group = block_group;
-	ac.ac_b_ex.fe_start = bit;
-	ac.ac_b_ex.fe_len = count;
-	ext4_mb_store_history(&ac);
+	if (ac) {
+		ac->ac_b_ex.fe_group = block_group;
+		ac->ac_b_ex.fe_start = bit;
+		ac->ac_b_ex.fe_len = count;
+		ext4_mb_store_history(ac);
+	}
 
 	if (metadata) {
 		/* blocks being freed are metadata. these blocks shouldn't
@@ -4522,8 +4591,7 @@ do_more:
 	}
 
 	spin_lock(sb_bgl_lock(sbi, block_group));
-	gdp->bg_free_blocks_count =
-		cpu_to_le16(le16_to_cpu(gdp->bg_free_blocks_count) + count);
+	le16_add_cpu(&gdp->bg_free_blocks_count, count);
 	gdp->bg_checksum = ext4_group_desc_csum(sbi, block_group, gdp);
 	spin_unlock(sb_bgl_lock(sbi, block_group));
 	percpu_counter_add(&sbi->s_freeblocks_counter, count);
@@ -4548,5 +4616,7 @@ do_more:
 error_return:
 	brelse(bitmap_bh);
 	ext4_std_error(sb, err);
+	if (ac)
+		kmem_cache_free(ext4_ac_cachep, ac);
 	return;
 }