1 files changed, 87 insertions, 126 deletions

diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 4415beeb0b6..f871677a798 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -46,22 +46,23 @@
  * The allocation request involve request for multiple number of blocks
  * near to the goal(block) value specified.
  *
- * During initialization phase of the allocator we decide to use the group
- * preallocation or inode preallocation depending on the size file. The
- * size of the file could be the resulting file size we would have after
- * allocation or the current file size which ever is larger. If the size is
- * less that sbi->s_mb_stream_request we select the group
- * preallocation. The default value of s_mb_stream_request is 16
- * blocks. This can also be tuned via
- * /proc/fs/ext4/<partition>/stream_req. The value is represented in terms
- * of number of blocks.
+ * During initialization phase of the allocator we decide to use the
+ * group preallocation or inode preallocation depending on the size of
+ * the file. The size of the file could be the resulting file size we
+ * would have after allocation, or the current file size, which ever
+ * is larger. If the size is less than sbi->s_mb_stream_request we
+ * select to use the group preallocation. The default value of
+ * s_mb_stream_request is 16 blocks. This can also be tuned via
+ * /sys/fs/ext4/<partition>/mb_stream_req. The value is represented in
+ * terms of number of blocks.
  *
  * The main motivation for having small file use group preallocation is to
- * ensure that we have small file closer in the disk.
+ * ensure that we have small files closer together on the disk.
  *
- * First stage the allocator looks at the inode prealloc list
- * ext4_inode_info->i_prealloc_list contain list of prealloc spaces for
- * this particular inode. The inode prealloc space is represented as:
+ * First stage the allocator looks at the inode prealloc list,
+ * ext4_inode_info->i_prealloc_list, which contains list of prealloc
+ * spaces for this particular inode. The inode prealloc space is
+ * represented as:
  *
  * pa_lstart -> the logical start block for this prealloc space
  * pa_pstart -> the physical start block for this prealloc space
@@ -121,29 +122,29 @@
  * list. In case of inode preallocation we follow a list of heuristics
  * based on file size. This can be found in ext4_mb_normalize_request. If
  * we are doing a group prealloc we try to normalize the request to
- * sbi->s_mb_group_prealloc. Default value of s_mb_group_prealloc is set to
+ * sbi->s_mb_group_prealloc. Default value of s_mb_group_prealloc is
  * 512 blocks. This can be tuned via
- * /proc/fs/ext4/<partition/group_prealloc. The value is represented in
+ * /sys/fs/ext4/<partition/mb_group_prealloc. The value is represented in
  * terms of number of blocks. If we have mounted the file system with -O
  * stripe=<value> option the group prealloc request is normalized to the
  * stripe value (sbi->s_stripe)
  *
- * The regular allocator(using the buddy cache) support few tunables.
+ * The regular allocator(using the buddy cache) supports few tunables.
  *
- * /proc/fs/ext4/<partition>/min_to_scan
- * /proc/fs/ext4/<partition>/max_to_scan
- * /proc/fs/ext4/<partition>/order2_req
+ * /sys/fs/ext4/<partition>/mb_min_to_scan
+ * /sys/fs/ext4/<partition>/mb_max_to_scan
+ * /sys/fs/ext4/<partition>/mb_order2_req
  *
- * The regular allocator use buddy scan only if the request len is power of
+ * The regular allocator uses buddy scan only if the request len is power of
  * 2 blocks and the order of allocation is >= sbi->s_mb_order2_reqs. The
  * value of s_mb_order2_reqs can be tuned via
- * /proc/fs/ext4/<partition>/order2_req.  If the request len is equal to
+ * /sys/fs/ext4/<partition>/mb_order2_req.  If the request len is equal to
  * stripe size (sbi->s_stripe), we try to search for contigous block in
- * stripe size. This should result in better allocation on RAID setup. If
- * not we search in the specific group using bitmap for best extents. The
- * tunable min_to_scan and max_to_scan controll the behaviour here.
+ * stripe size. This should result in better allocation on RAID setups. If
+ * not, we search in the specific group using bitmap for best extents. The
+ * tunable min_to_scan and max_to_scan control the behaviour here.
  * min_to_scan indicate how long the mballoc __must__ look for a best
- * extent and max_to_scanindicate how long the mballoc __can__ look for a
+ * extent and max_to_scan indicates how long the mballoc __can__ look for a
  * best extent in the found extents. Searching for the blocks starts with
  * the group specified as the goal value in allocation context via
  * ac_g_ex. Each group is first checked based on the criteria whether it
@@ -337,8 +338,6 @@ static void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap,
 					ext4_group_t group);
 static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap,
 						ext4_group_t group);
-static int ext4_mb_init_per_dev_proc(struct super_block *sb);
-static int ext4_mb_destroy_per_dev_proc(struct super_block *sb);
 static void release_blocks_on_commit(journal_t *journal, transaction_t *txn);
 
 
@@ -1447,7 +1446,7 @@ static void ext4_mb_measure_extent(struct ext4_allocation_context *ac,
 	struct ext4_free_extent *gex = &ac->ac_g_ex;
 
 	BUG_ON(ex->fe_len <= 0);
-	BUG_ON(ex->fe_len >= EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
+	BUG_ON(ex->fe_len > EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
 	BUG_ON(ex->fe_start >= EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
 	BUG_ON(ac->ac_status != AC_STATUS_CONTINUE);
 
@@ -1726,6 +1725,7 @@ static int ext4_mb_good_group(struct ext4_allocation_context *ac,
 {
 	unsigned free, fragments;
 	unsigned i, bits;
+	int flex_size = ext4_flex_bg_size(EXT4_SB(ac->ac_sb));
 	struct ext4_group_desc *desc;
 	struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group);
 
@@ -1747,6 +1747,12 @@ static int ext4_mb_good_group(struct ext4_allocation_context *ac,
 		if (desc->bg_flags & cpu_to_le16(EXT4_BG_BLOCK_UNINIT))
 			return 0;
 
+		/* Avoid using the first bg of a flexgroup for data files */
+		if ((ac->ac_flags & EXT4_MB_HINT_DATA) &&
+		    (flex_size >= EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME) &&
+		    ((group % flex_size) == 0))
+			return 0;
+
 		bits = ac->ac_sb->s_blocksize_bits + 1;
 		for (i = ac->ac_2order; i <= bits; i++)
 			if (grp->bb_counters[i] > 0)
@@ -1971,7 +1977,7 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 	/*
 	 * We search using buddy data only if the order of the request
 	 * is greater than equal to the sbi_s_mb_order2_reqs
-	 * You can tune it via /proc/fs/ext4/<partition>/order2_req
+	 * You can tune it via /sys/fs/ext4/<partition>/mb_order2_req
 	 */
 	if (i >= sbi->s_mb_order2_reqs) {
 		/*
@@ -2693,7 +2699,7 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery)
 	i = (sb->s_blocksize_bits + 2) * sizeof(unsigned int);
 	sbi->s_mb_maxs = kmalloc(i, GFP_KERNEL);
 	if (sbi->s_mb_maxs == NULL) {
-		kfree(sbi->s_mb_maxs);
+		kfree(sbi->s_mb_offsets);
 		return -ENOMEM;
 	}
 
@@ -2746,7 +2752,6 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery)
 		spin_lock_init(&lg->lg_prealloc_lock);
 	}
 
-	ext4_mb_init_per_dev_proc(sb);
 	ext4_mb_history_init(sb);
 
 	if (sbi->s_journal)
@@ -2829,7 +2834,6 @@ int ext4_mb_release(struct super_block *sb)
 
 	free_percpu(sbi->s_locality_groups);
 	ext4_mb_history_release(sb);
-	ext4_mb_destroy_per_dev_proc(sb);
 
 	return 0;
 }
@@ -2890,62 +2894,6 @@ static void release_blocks_on_commit(journal_t *journal, transaction_t *txn)
 	mb_debug("freed %u blocks in %u structures\n", count, count2);
 }
 
-#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"
-#define EXT4_MB_ORDER2_REQ		"order2_req"
-#define EXT4_MB_STREAM_REQ		"stream_req"
-#define EXT4_MB_GROUP_PREALLOC		"group_prealloc"
-
-static int ext4_mb_init_per_dev_proc(struct super_block *sb)
-{
-#ifdef CONFIG_PROC_FS
-	mode_t mode = S_IFREG | S_IRUGO | S_IWUSR;
-	struct ext4_sb_info *sbi = EXT4_SB(sb);
-	struct proc_dir_entry *proc;
-
-	if (sbi->s_proc == NULL)
-		return -EINVAL;
-
-	EXT4_PROC_HANDLER(EXT4_MB_STATS_NAME, mb_stats);
-	EXT4_PROC_HANDLER(EXT4_MB_MAX_TO_SCAN_NAME, mb_max_to_scan);
-	EXT4_PROC_HANDLER(EXT4_MB_MIN_TO_SCAN_NAME, mb_min_to_scan);
-	EXT4_PROC_HANDLER(EXT4_MB_ORDER2_REQ, mb_order2_reqs);
-	EXT4_PROC_HANDLER(EXT4_MB_STREAM_REQ, mb_stream_request);
-	EXT4_PROC_HANDLER(EXT4_MB_GROUP_PREALLOC, mb_group_prealloc);
-	return 0;
-
-err_out:
-	remove_proc_entry(EXT4_MB_GROUP_PREALLOC, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_STREAM_REQ, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_ORDER2_REQ, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_MIN_TO_SCAN_NAME, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_MAX_TO_SCAN_NAME, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_STATS_NAME, sbi->s_proc);
-	return -ENOMEM;
-#else
-	return 0;
-#endif
-}
-
-static int ext4_mb_destroy_per_dev_proc(struct super_block *sb)
-{
-#ifdef CONFIG_PROC_FS
-	struct ext4_sb_info *sbi = EXT4_SB(sb);
-
-	if (sbi->s_proc == NULL)
-		return -EINVAL;
-
-	remove_proc_entry(EXT4_MB_GROUP_PREALLOC, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_STREAM_REQ, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_ORDER2_REQ, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_MIN_TO_SCAN_NAME, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_MAX_TO_SCAN_NAME, sbi->s_proc);
-	remove_proc_entry(EXT4_MB_STATS_NAME, sbi->s_proc);
-#endif
-	return 0;
-}
-
 int __init init_ext4_mballoc(void)
 {
 	ext4_pspace_cachep =
@@ -3086,16 +3034,18 @@ ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
 	if (!(ac->ac_flags & EXT4_MB_DELALLOC_RESERVED))
 		/* release all the reserved blocks if non delalloc */
 		percpu_counter_sub(&sbi->s_dirtyblocks_counter, reserv_blks);
-	else
+	else {
 		percpu_counter_sub(&sbi->s_dirtyblocks_counter,
 						ac->ac_b_ex.fe_len);
+		/* convert reserved quota blocks to real quota blocks */
+		vfs_dq_claim_block(ac->ac_inode, ac->ac_b_ex.fe_len);
+	}
 
 	if (sbi->s_log_groups_per_flex) {
 		ext4_group_t flex_group = ext4_flex_group(sbi,
 							  ac->ac_b_ex.fe_group);
-		spin_lock(sb_bgl_lock(sbi, flex_group));
-		sbi->s_flex_groups[flex_group].free_blocks -= ac->ac_b_ex.fe_len;
-		spin_unlock(sb_bgl_lock(sbi, flex_group));
+		atomic_sub(ac->ac_b_ex.fe_len,
+			   &sbi->s_flex_groups[flex_group].free_blocks);
 	}
 
 	err = ext4_handle_dirty_metadata(handle, NULL, bitmap_bh);
@@ -3113,7 +3063,7 @@ out_err:
  * here we normalize request for locality group
  * Group request are normalized to s_strip size if we set the same via mount
  * option. If not we set it to s_mb_group_prealloc which can be configured via
- * /proc/fs/ext4/<partition>/group_prealloc
+ * /sys/fs/ext4/<partition>/mb_group_prealloc
  *
  * XXX: should we try to preallocate more than the group has now?
  */
@@ -3292,7 +3242,7 @@ ext4_mb_normalize_request(struct ext4_allocation_context *ac,
 	}
 	BUG_ON(start + size <= ac->ac_o_ex.fe_logical &&
 			start > ac->ac_o_ex.fe_logical);
-	BUG_ON(size <= 0 || size >= EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
+	BUG_ON(size <= 0 || size > EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
 
 	/* now prepare goal request */
 
@@ -3589,6 +3539,7 @@ static void ext4_mb_put_pa(struct ext4_allocation_context *ac,
 			struct super_block *sb, struct ext4_prealloc_space *pa)
 {
 	ext4_group_t grp;
+	ext4_fsblk_t grp_blk;
 
 	if (!atomic_dec_and_test(&pa->pa_count) || pa->pa_free != 0)
 		return;
@@ -3603,8 +3554,15 @@ static void ext4_mb_put_pa(struct ext4_allocation_context *ac,
 	pa->pa_deleted = 1;
 	spin_unlock(&pa->pa_lock);
 
-	/* -1 is to protect from crossing allocation group */
-	ext4_get_group_no_and_offset(sb, pa->pa_pstart - 1, &grp, NULL);
+	grp_blk = pa->pa_pstart;
+	/* 
+	 * If doing group-based preallocation, pa_pstart may be in the
+	 * next group when pa is used up
+	 */
+	if (pa->pa_type == MB_GROUP_PA)
+		grp_blk--;
+
+	ext4_get_group_no_and_offset(sb, grp_blk, &grp, NULL);
 
 	/*
 	 * possible race:
@@ -3696,7 +3654,7 @@ ext4_mb_new_inode_pa(struct ext4_allocation_context *ac)
 	INIT_LIST_HEAD(&pa->pa_inode_list);
 	INIT_LIST_HEAD(&pa->pa_group_list);
 	pa->pa_deleted = 0;
-	pa->pa_linear = 0;
+	pa->pa_type = MB_INODE_PA;
 
 	mb_debug("new inode pa %p: %llu/%u for %u\n", pa,
 			pa->pa_pstart, pa->pa_len, pa->pa_lstart);
@@ -3759,7 +3717,7 @@ ext4_mb_new_group_pa(struct ext4_allocation_context *ac)
 	INIT_LIST_HEAD(&pa->pa_inode_list);
 	INIT_LIST_HEAD(&pa->pa_group_list);
 	pa->pa_deleted = 0;
-	pa->pa_linear = 1;
+	pa->pa_type = MB_GROUP_PA;
 
 	mb_debug("new group pa %p: %llu/%u for %u\n", pa,
 		 pa->pa_pstart, pa->pa_len, pa->pa_lstart);
@@ -4013,7 +3971,7 @@ repeat:
 		list_del_rcu(&pa->pa_inode_list);
 		spin_unlock(pa->pa_obj_lock);
 
-		if (pa->pa_linear)
+		if (pa->pa_type == MB_GROUP_PA)
 			ext4_mb_release_group_pa(&e4b, pa, ac);
 		else
 			ext4_mb_release_inode_pa(&e4b, bitmap_bh, pa, ac);
@@ -4113,7 +4071,7 @@ repeat:
 	spin_unlock(&ei->i_prealloc_lock);
 
 	list_for_each_entry_safe(pa, tmp, &list, u.pa_tmp_list) {
-		BUG_ON(pa->pa_linear != 0);
+		BUG_ON(pa->pa_type != MB_INODE_PA);
 		ext4_get_group_no_and_offset(sb, pa->pa_pstart, &group, NULL);
 
 		err = ext4_mb_load_buddy(sb, group, &e4b);
@@ -4224,7 +4182,7 @@ static inline void ext4_mb_show_ac(struct ext4_allocation_context *ac)
  * file is determined by the current size or the resulting size after
  * allocation which ever is larger
  *
- * One can tune this size via /proc/fs/ext4/<partition>/stream_req
+ * One can tune this size via /sys/fs/ext4/<partition>/mb_stream_req
  */
 static void ext4_mb_group_or_file(struct ext4_allocation_context *ac)
 {
@@ -4365,7 +4323,7 @@ ext4_mb_discard_lg_preallocations(struct super_block *sb,
 			continue;
 		}
 		/* only lg prealloc space */
-		BUG_ON(!pa->pa_linear);
+		BUG_ON(pa->pa_type != MB_GROUP_PA);
 
 		/* seems this one can be freed ... */
 		pa->pa_deleted = 1;
@@ -4434,7 +4392,7 @@ static void ext4_mb_add_n_trim(struct ext4_allocation_context *ac)
 						pa_inode_list) {
 		spin_lock(&tmp_pa->pa_lock);
 		if (tmp_pa->pa_deleted) {
-			spin_unlock(&pa->pa_lock);
+			spin_unlock(&tmp_pa->pa_lock);
 			continue;
 		}
 		if (!added && pa->pa_free < tmp_pa->pa_free) {
@@ -4471,7 +4429,7 @@ static int ext4_mb_release_context(struct ext4_allocation_context *ac)
 {
 	struct ext4_prealloc_space *pa = ac->ac_pa;
 	if (pa) {
-		if (pa->pa_linear) {
+		if (pa->pa_type == MB_GROUP_PA) {
 			/* see comment in ext4_mb_use_group_pa() */
 			spin_lock(&pa->pa_lock);
 			pa->pa_pstart += ac->ac_b_ex.fe_len;
@@ -4491,7 +4449,7 @@ static int ext4_mb_release_context(struct ext4_allocation_context *ac)
 		 * doesn't grow big.  We need to release
 		 * alloc_semp before calling ext4_mb_add_n_trim()
 		 */
-		if (pa->pa_linear && likely(pa->pa_free)) {
+		if ((pa->pa_type == MB_GROUP_PA) && likely(pa->pa_free)) {
 			spin_lock(pa->pa_obj_lock);
 			list_del_rcu(&pa->pa_inode_list);
 			spin_unlock(pa->pa_obj_lock);
@@ -4539,7 +4497,7 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
 	struct ext4_sb_info *sbi;
 	struct super_block *sb;
 	ext4_fsblk_t block = 0;
-	unsigned int inquota;
+	unsigned int inquota = 0;
 	unsigned int reserv_blks = 0;
 
 	sb = ar->inode->i_sb;
@@ -4557,9 +4515,17 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
 		   (unsigned long long) ar->pleft,
 		   (unsigned long long) ar->pright);
 
-	if (!EXT4_I(ar->inode)->i_delalloc_reserved_flag) {
-		/*
-		 * With delalloc we already reserved the blocks
+	/*
+	 * For delayed allocation, we could skip the ENOSPC and
+	 * EDQUOT check, as blocks and quotas have been already
+	 * reserved when data being copied into pagecache.
+	 */
+	if (EXT4_I(ar->inode)->i_delalloc_reserved_flag)
+		ar->flags |= EXT4_MB_DELALLOC_RESERVED;
+	else {
+		/* Without delayed allocation we need to verify
+		 * there is enough free blocks to do block allocation
+		 * and verify allocation doesn't exceed the quota limits.
 		 */
 		while (ar->len && ext4_claim_free_blocks(sbi, ar->len)) {
 			/* let others to free the space */
@@ -4571,19 +4537,16 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
 			return 0;
 		}
 		reserv_blks = ar->len;
+		while (ar->len && vfs_dq_alloc_block(ar->inode, ar->len)) {
+			ar->flags |= EXT4_MB_HINT_NOPREALLOC;
+			ar->len--;
+		}
+		inquota = ar->len;
+		if (ar->len == 0) {
+			*errp = -EDQUOT;
+			goto out3;
+		}
 	}
-	while (ar->len && DQUOT_ALLOC_BLOCK(ar->inode, ar->len)) {
-		ar->flags |= EXT4_MB_HINT_NOPREALLOC;
-		ar->len--;
-	}
-	if (ar->len == 0) {
-		*errp = -EDQUOT;
-		goto out3;
-	}
-	inquota = ar->len;
-
-	if (EXT4_I(ar->inode)->i_delalloc_reserved_flag)
-		ar->flags |= EXT4_MB_DELALLOC_RESERVED;
 
 	ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
 	if (!ac) {
@@ -4649,8 +4612,8 @@ repeat:
 out2:
 	kmem_cache_free(ext4_ac_cachep, ac);
 out1:
-	if (ar->len < inquota)
-		DQUOT_FREE_BLOCK(ar->inode, inquota - ar->len);
+	if (inquota && ar->len < inquota)
+		vfs_dq_free_block(ar->inode, inquota - ar->len);
 out3:
 	if (!ar->len) {
 		if (!EXT4_I(ar->inode)->i_delalloc_reserved_flag)
@@ -4923,9 +4886,7 @@ do_more:
 
 	if (sbi->s_log_groups_per_flex) {
 		ext4_group_t flex_group = ext4_flex_group(sbi, block_group);
-		spin_lock(sb_bgl_lock(sbi, flex_group));
-		sbi->s_flex_groups[flex_group].free_blocks += count;
-		spin_unlock(sb_bgl_lock(sbi, flex_group));
+		atomic_add(count, &sbi->s_flex_groups[flex_group].free_blocks);
 	}
 
 	ext4_mb_release_desc(&e4b);