1 files changed, 101 insertions, 35 deletions

diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index a9ff5e5137c..2dcb936be90 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -722,6 +722,7 @@ void ext4_mb_generate_buddy(struct super_block *sb,
 				void *buddy, void *bitmap, ext4_group_t group)
 {
 	struct ext4_group_info *grp = ext4_get_group_info(sb, group);
+	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	ext4_grpblk_t max = EXT4_CLUSTERS_PER_GROUP(sb);
 	ext4_grpblk_t i = 0;
 	ext4_grpblk_t first;
@@ -751,13 +752,18 @@ void ext4_mb_generate_buddy(struct super_block *sb,
 
 	if (free != grp->bb_free) {
 		ext4_grp_locked_error(sb, group, 0, 0,
-				      "%u clusters in bitmap, %u in gd",
+				      "block bitmap and bg descriptor "
+				      "inconsistent: %u vs %u free clusters",
 				      free, grp->bb_free);
 		/*
-		 * If we intent to continue, we consider group descritor
+		 * If we intend to continue, we consider group descriptor
 		 * corrupt and update bb_free using bitmap value
 		 */
 		grp->bb_free = free;
+		if (!EXT4_MB_GRP_BBITMAP_CORRUPT(grp))
+			percpu_counter_sub(&sbi->s_freeclusters_counter,
+					   grp->bb_free);
+		set_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &grp->bb_state);
 	}
 	mb_set_largest_free_order(sb, grp);
 
@@ -987,7 +993,7 @@ static int ext4_mb_get_buddy_page_lock(struct super_block *sb,
 	poff = block % blocks_per_page;
 	page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
 	if (!page)
-		return -EIO;
+		return -ENOMEM;
 	BUG_ON(page->mapping != inode->i_mapping);
 	e4b->bd_bitmap_page = page;
 	e4b->bd_bitmap = page_address(page) + (poff * sb->s_blocksize);
@@ -1001,7 +1007,7 @@ static int ext4_mb_get_buddy_page_lock(struct super_block *sb,
 	pnum = block / blocks_per_page;
 	page = find_or_create_page(inode->i_mapping, pnum, GFP_NOFS);
 	if (!page)
-		return -EIO;
+		return -ENOMEM;
 	BUG_ON(page->mapping != inode->i_mapping);
 	e4b->bd_buddy_page = page;
 	return 0;
@@ -1042,6 +1048,8 @@ int ext4_mb_init_group(struct super_block *sb, ext4_group_t group)
 	 * allocating. If we are looking at the buddy cache we would
 	 * have taken a reference using ext4_mb_load_buddy and that
 	 * would have pinned buddy page to page cache.
+	 * The call to ext4_mb_get_buddy_page_lock will mark the
+	 * page accessed.
 	 */
 	ret = ext4_mb_get_buddy_page_lock(sb, group, &e4b);
 	if (ret || !EXT4_MB_GRP_NEED_INIT(this_grp)) {
@@ -1060,7 +1068,6 @@ int ext4_mb_init_group(struct super_block *sb, ext4_group_t group)
 		ret = -EIO;
 		goto err;
 	}
-	mark_page_accessed(page);
 
 	if (e4b.bd_buddy_page == NULL) {
 		/*
@@ -1080,7 +1087,6 @@ int ext4_mb_init_group(struct super_block *sb, ext4_group_t group)
 		ret = -EIO;
 		goto err;
 	}
-	mark_page_accessed(page);
 err:
 	ext4_mb_put_buddy_page_lock(&e4b);
 	return ret;
@@ -1139,7 +1145,7 @@ ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
 
 	/* we could use find_or_create_page(), but it locks page
 	 * what we'd like to avoid in fast path ... */
-	page = find_get_page(inode->i_mapping, pnum);
+	page = find_get_page_flags(inode->i_mapping, pnum, FGP_ACCESSED);
 	if (page == NULL || !PageUptodate(page)) {
 		if (page)
 			/*
@@ -1166,19 +1172,24 @@ ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
 			unlock_page(page);
 		}
 	}
-	if (page == NULL || !PageUptodate(page)) {
+	if (page == NULL) {
+		ret = -ENOMEM;
+		goto err;
+	}
+	if (!PageUptodate(page)) {
 		ret = -EIO;
 		goto err;
 	}
+
+	/* Pages marked accessed already */
 	e4b->bd_bitmap_page = page;
 	e4b->bd_bitmap = page_address(page) + (poff * sb->s_blocksize);
-	mark_page_accessed(page);
 
 	block++;
 	pnum = block / blocks_per_page;
 	poff = block % blocks_per_page;
 
-	page = find_get_page(inode->i_mapping, pnum);
+	page = find_get_page_flags(inode->i_mapping, pnum, FGP_ACCESSED);
 	if (page == NULL || !PageUptodate(page)) {
 		if (page)
 			page_cache_release(page);
@@ -1195,13 +1206,18 @@ ext4_mb_load_buddy(struct super_block *sb, ext4_group_t group,
 			unlock_page(page);
 		}
 	}
-	if (page == NULL || !PageUptodate(page)) {
+	if (page == NULL) {
+		ret = -ENOMEM;
+		goto err;
+	}
+	if (!PageUptodate(page)) {
 		ret = -EIO;
 		goto err;
 	}
+
+	/* Pages marked accessed already */
 	e4b->bd_buddy_page = page;
 	e4b->bd_buddy = page_address(page) + (poff * sb->s_blocksize);
-	mark_page_accessed(page);
 
 	BUG_ON(e4b->bd_bitmap_page == NULL);
 	BUG_ON(e4b->bd_buddy_page == NULL);
@@ -1398,6 +1414,10 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b,
 
 	BUG_ON(last >= (sb->s_blocksize << 3));
 	assert_spin_locked(ext4_group_lock_ptr(sb, e4b->bd_group));
+	/* Don't bother if the block group is corrupt. */
+	if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info)))
+		return;
+
 	mb_check_buddy(e4b);
 	mb_free_blocks_double(inode, e4b, first, count);
 
@@ -1415,6 +1435,7 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b,
 		right_is_free = !mb_test_bit(last + 1, e4b->bd_bitmap);
 
 	if (unlikely(block != -1)) {
+		struct ext4_sb_info *sbi = EXT4_SB(sb);
 		ext4_fsblk_t blocknr;
 
 		blocknr = ext4_group_first_block_no(sb, e4b->bd_group);
@@ -1423,7 +1444,14 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b,
 				      inode ? inode->i_ino : 0,
 				      blocknr,
 				      "freeing already freed block "
-				      "(bit %u)", block);
+				      "(bit %u); block bitmap corrupt.",
+				      block);
+		if (!EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info))
+			percpu_counter_sub(&sbi->s_freeclusters_counter,
+					   e4b->bd_info->bb_free);
+		/* Mark the block group as corrupt. */
+		set_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT,
+			&e4b->bd_info->bb_state);
 		mb_regenerate_buddy(e4b);
 		goto done;
 	}
@@ -1790,9 +1818,15 @@ int ext4_mb_find_by_goal(struct ext4_allocation_context *ac,
 	if (err)
 		return err;
 
+	if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(e4b->bd_info))) {
+		ext4_mb_unload_buddy(e4b);
+		return 0;
+	}
+
 	ext4_lock_group(ac->ac_sb, group);
 	max = mb_find_extent(e4b, ac->ac_g_ex.fe_start,
 			     ac->ac_g_ex.fe_len, &ex);
+	ex.fe_logical = 0xDEADFA11; /* debug value */
 
 	if (max >= ac->ac_g_ex.fe_len && ac->ac_g_ex.fe_len == sbi->s_stripe) {
 		ext4_fsblk_t start;
@@ -1921,7 +1955,7 @@ void ext4_mb_complex_scan_group(struct ext4_allocation_context *ac,
 			 */
 			break;
 		}
-
+		ex.fe_logical = 0xDEADC0DE; /* debug value */
 		ext4_mb_measure_extent(ac, &ex, e4b);
 
 		i += ex.fe_len;
@@ -1962,6 +1996,7 @@ void ext4_mb_scan_aligned(struct ext4_allocation_context *ac,
 			max = mb_find_extent(e4b, i, sbi->s_stripe, &ex);
 			if (max >= sbi->s_stripe) {
 				ac->ac_found++;
+				ex.fe_logical = 0xDEADF00D; /* debug value */
 				ac->ac_b_ex = ex;
 				ext4_mb_use_best_found(ac, e4b);
 				break;
@@ -1987,6 +2022,9 @@ static int ext4_mb_good_group(struct ext4_allocation_context *ac,
 	if (cr <= 2 && free < ac->ac_g_ex.fe_len)
 		return 0;
 
+	if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(grp)))
+		return 0;
+
 	/* We only do this if the grp has never been initialized */
 	if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) {
 		int ret = ext4_mb_init_group(ac->ac_sb, group);
@@ -2589,7 +2627,7 @@ int ext4_mb_init(struct super_block *sb)
 	sbi->s_locality_groups = alloc_percpu(struct ext4_locality_group);
 	if (sbi->s_locality_groups == NULL) {
 		ret = -ENOMEM;
-		goto out_free_groupinfo_slab;
+		goto out;
 	}
 	for_each_possible_cpu(i) {
 		struct ext4_locality_group *lg;
@@ -2614,8 +2652,6 @@ int ext4_mb_init(struct super_block *sb)
 out_free_locality_groups:
 	free_percpu(sbi->s_locality_groups);
 	sbi->s_locality_groups = NULL;
-out_free_groupinfo_slab:
-	ext4_groupinfo_destroy_slabs();
 out:
 	kfree(sbi->s_mb_offsets);
 	sbi->s_mb_offsets = NULL;
@@ -2848,6 +2884,7 @@ ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
 	if (!bitmap_bh)
 		goto out_err;
 
+	BUFFER_TRACE(bitmap_bh, "getting write access");
 	err = ext4_journal_get_write_access(handle, bitmap_bh);
 	if (err)
 		goto out_err;
@@ -2860,6 +2897,7 @@ ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
 	ext4_debug("using block group %u(%d)\n", ac->ac_b_ex.fe_group,
 			ext4_free_group_clusters(sb, gdp));
 
+	BUFFER_TRACE(gdp_bh, "get_write_access");
 	err = ext4_journal_get_write_access(handle, gdp_bh);
 	if (err)
 		goto out_err;
@@ -3117,7 +3155,7 @@ 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_CLUSTERS_PER_GROUP(ac->ac_sb));
+	BUG_ON(size <= 0 || size > EXT4_BLOCKS_PER_GROUP(ac->ac_sb));
 
 	/* now prepare goal request */
 
@@ -3424,6 +3462,9 @@ static void ext4_mb_pa_callback(struct rcu_head *head)
 {
 	struct ext4_prealloc_space *pa;
 	pa = container_of(head, struct ext4_prealloc_space, u.pa_rcu);
+
+	BUG_ON(atomic_read(&pa->pa_count));
+	BUG_ON(pa->pa_deleted == 0);
 	kmem_cache_free(ext4_pspace_cachep, pa);
 }
 
@@ -3437,11 +3478,13 @@ static void ext4_mb_put_pa(struct ext4_allocation_context *ac,
 	ext4_group_t grp;
 	ext4_fsblk_t grp_blk;
 
-	if (!atomic_dec_and_test(&pa->pa_count) || pa->pa_free != 0)
-		return;
-
 	/* in this short window concurrent discard can set pa_deleted */
 	spin_lock(&pa->pa_lock);
+	if (!atomic_dec_and_test(&pa->pa_count) || pa->pa_free != 0) {
+		spin_unlock(&pa->pa_lock);
+		return;
+	}
+
 	if (pa->pa_deleted == 1) {
 		spin_unlock(&pa->pa_lock);
 		return;
@@ -3983,8 +4026,7 @@ static void ext4_mb_show_ac(struct ext4_allocation_context *ac)
 			(unsigned long)ac->ac_b_ex.fe_len,
 			(unsigned long)ac->ac_b_ex.fe_logical,
 			(int)ac->ac_criteria);
-	ext4_msg(ac->ac_sb, KERN_ERR, "%lu scanned, %d found",
-		 ac->ac_ex_scanned, ac->ac_found);
+	ext4_msg(ac->ac_sb, KERN_ERR, "%d found", ac->ac_found);
 	ext4_msg(ac->ac_sb, KERN_ERR, "groups: ");
 	ngroups = ext4_get_groups_count(sb);
 	for (i = 0; i < ngroups; i++) {
@@ -4103,7 +4145,7 @@ ext4_mb_initialize_context(struct ext4_allocation_context *ac,
 	ext4_get_group_no_and_offset(sb, goal, &group, &block);
 
 	/* set up allocation goals */
-	ac->ac_b_ex.fe_logical = ar->logical & ~(sbi->s_cluster_ratio - 1);
+	ac->ac_b_ex.fe_logical = EXT4_LBLK_CMASK(sbi, ar->logical);
 	ac->ac_status = AC_STATUS_CONTINUE;
 	ac->ac_sb = sb;
 	ac->ac_inode = ar->inode;
@@ -4585,6 +4627,7 @@ void ext4_free_blocks(handle_t *handle, struct inode *inode,
 	struct buffer_head *gd_bh;
 	ext4_group_t block_group;
 	struct ext4_sb_info *sbi;
+	struct ext4_inode_info *ei = EXT4_I(inode);
 	struct ext4_buddy e4b;
 	unsigned int count_clusters;
 	int err = 0;
@@ -4644,7 +4687,7 @@ void ext4_free_blocks(handle_t *handle, struct inode *inode,
 	 * blocks at the beginning or the end unless we are explicitly
 	 * requested to avoid doing so.
 	 */
-	overflow = block & (sbi->s_cluster_ratio - 1);
+	overflow = EXT4_PBLK_COFF(sbi, block);
 	if (overflow) {
 		if (flags & EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER) {
 			overflow = sbi->s_cluster_ratio - overflow;
@@ -4658,7 +4701,7 @@ void ext4_free_blocks(handle_t *handle, struct inode *inode,
 			count += overflow;
 		}
 	}
-	overflow = count & (sbi->s_cluster_ratio - 1);
+	overflow = EXT4_LBLK_COFF(sbi, count);
 	if (overflow) {
 		if (flags & EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER) {
 			if (count > overflow)
@@ -4673,6 +4716,10 @@ do_more:
 	overflow = 0;
 	ext4_get_group_no_and_offset(sb, block, &block_group, &bit);
 
+	if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(
+			ext4_get_group_info(sb, block_group))))
+		return;
+
 	/*
 	 * Check to see if we are freeing blocks across a group
 	 * boundary.
@@ -4740,11 +4787,16 @@ do_more:
 		 * blocks being freed are metadata. these blocks shouldn't
 		 * be used until this transaction is committed
 		 */
+	retry:
 		new_entry = kmem_cache_alloc(ext4_free_data_cachep, GFP_NOFS);
 		if (!new_entry) {
-			ext4_mb_unload_buddy(&e4b);
-			err = -ENOMEM;
-			goto error_return;
+			/*
+			 * We use a retry loop because
+			 * ext4_free_blocks() is not allowed to fail.
+			 */
+			cond_resched();
+			congestion_wait(BLK_RW_ASYNC, HZ/50);
+			goto retry;
 		}
 		new_entry->efd_start_cluster = bit;
 		new_entry->efd_group = block_group;
@@ -4766,8 +4818,8 @@ do_more:
 					 " group:%d block:%d count:%lu failed"
 					 " with %d", block_group, bit, count,
 					 err);
-		}
-
+		} else
+			EXT4_MB_GRP_CLEAR_TRIMMED(e4b.bd_info);
 
 		ext4_lock_group(sb, block_group);
 		mb_clear_bits(bitmap_bh->b_data, bit, count_clusters);
@@ -4779,7 +4831,6 @@ do_more:
 	ext4_block_bitmap_csum_set(sb, block_group, gdp, bitmap_bh);
 	ext4_group_desc_csum_set(sb, block_group, gdp);
 	ext4_unlock_group(sb, block_group);
-	percpu_counter_add(&sbi->s_freeclusters_counter, count_clusters);
 
 	if (sbi->s_log_groups_per_flex) {
 		ext4_group_t flex_group = ext4_flex_group(sbi, block_group);
@@ -4787,10 +4838,23 @@ do_more:
 			     &sbi->s_flex_groups[flex_group].free_clusters);
 	}
 
-	ext4_mb_unload_buddy(&e4b);
-
-	if (!(flags & EXT4_FREE_BLOCKS_NO_QUOT_UPDATE))
+	if (flags & EXT4_FREE_BLOCKS_RESERVE && ei->i_reserved_data_blocks) {
+		percpu_counter_add(&sbi->s_dirtyclusters_counter,
+				   count_clusters);
+		spin_lock(&ei->i_block_reservation_lock);
+		if (flags & EXT4_FREE_BLOCKS_METADATA)
+			ei->i_reserved_meta_blocks += count_clusters;
+		else
+			ei->i_reserved_data_blocks += count_clusters;
+		spin_unlock(&ei->i_block_reservation_lock);
+		if (!(flags & EXT4_FREE_BLOCKS_NO_QUOT_UPDATE))
+			dquot_reclaim_block(inode,
+					EXT4_C2B(sbi, count_clusters));
+	} else if (!(flags & EXT4_FREE_BLOCKS_NO_QUOT_UPDATE))
 		dquot_free_block(inode, EXT4_C2B(sbi, count_clusters));
+	percpu_counter_add(&sbi->s_freeclusters_counter, count_clusters);
+
+	ext4_mb_unload_buddy(&e4b);
 
 	/* We dirtied the bitmap block */
 	BUFFER_TRACE(bitmap_bh, "dirtied bitmap block");
@@ -4962,6 +5026,8 @@ error_return:
  */
 static int ext4_trim_extent(struct super_block *sb, int start, int count,
 			     ext4_group_t group, struct ext4_buddy *e4b)
+__releases(bitlock)
+__acquires(bitlock)
 {
 	struct ext4_free_extent ex;
 	int ret = 0;