ext4: fix reserved cluster accounting at delayed write time

The code in ext4_da_map_blocks sometimes reserves space for more
delayed allocated clusters than it should, resulting in premature
ENOSPC, exceeded quota, and inaccurate free space reporting.

Fix this by checking for written and unwritten blocks shared in the
same cluster with the newly delayed allocated block.  A cluster
reservation should not be made for a cluster for which physical space
has already been allocated.

Signed-off-by: Eric Whitney <enwlinux@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

5 files changed, 206 insertions, 18 deletions

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index fc0f41dbf90b..d85fd5c8a2c4 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -3155,6 +3155,7 @@ extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
 				struct inode *inode2, ext4_lblk_t lblk1,
 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
 			     int mark_unwritten,int *err);
+extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
 
 /* move_extent.c */
 extern void ext4_double_down_write_data_sem(struct inode *first,
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 95796f00e4e6..26481e543312 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -5930,3 +5930,82 @@ ext4_swap_extents(handle_t *handle, struct inode *inode1,
 	}
 	return replaced_count;
 }
+
+/*
+ * ext4_clu_mapped - determine whether any block in a logical cluster has
+ *                   been mapped to a physical cluster
+ *
+ * @inode - file containing the logical cluster
+ * @lclu - logical cluster of interest
+ *
+ * Returns 1 if any block in the logical cluster is mapped, signifying
+ * that a physical cluster has been allocated for it.  Otherwise,
+ * returns 0.  Can also return negative error codes.  Derived from
+ * ext4_ext_map_blocks().
+ */
+int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu)
+{
+	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
+	struct ext4_ext_path *path;
+	int depth, mapped = 0, err = 0;
+	struct ext4_extent *extent;
+	ext4_lblk_t first_lblk, first_lclu, last_lclu;
+
+	/* search for the extent closest to the first block in the cluster */
+	path = ext4_find_extent(inode, EXT4_C2B(sbi, lclu), NULL, 0);
+	if (IS_ERR(path)) {
+		err = PTR_ERR(path);
+		path = NULL;
+		goto out;
+	}
+
+	depth = ext_depth(inode);
+
+	/*
+	 * A consistent leaf must not be empty.  This situation is possible,
+	 * though, _during_ tree modification, and it's why an assert can't
+	 * be put in ext4_find_extent().
+	 */
+	if (unlikely(path[depth].p_ext == NULL && depth != 0)) {
+		EXT4_ERROR_INODE(inode,
+		    "bad extent address - lblock: %lu, depth: %d, pblock: %lld",
+				 (unsigned long) EXT4_C2B(sbi, lclu),
+				 depth, path[depth].p_block);
+		err = -EFSCORRUPTED;
+		goto out;
+	}
+
+	extent = path[depth].p_ext;
+
+	/* can't be mapped if the extent tree is empty */
+	if (extent == NULL)
+		goto out;
+
+	first_lblk = le32_to_cpu(extent->ee_block);
+	first_lclu = EXT4_B2C(sbi, first_lblk);
+
+	/*
+	 * Three possible outcomes at this point - found extent spanning
+	 * the target cluster, to the left of the target cluster, or to the
+	 * right of the target cluster.  The first two cases are handled here.
+	 * The last case indicates the target cluster is not mapped.
+	 */
+	if (lclu >= first_lclu) {
+		last_lclu = EXT4_B2C(sbi, first_lblk +
+				     ext4_ext_get_actual_len(extent) - 1);
+		if (lclu <= last_lclu) {
+			mapped = 1;
+		} else {
+			first_lblk = ext4_ext_next_allocated_block(path);
+			first_lclu = EXT4_B2C(sbi, first_lblk);
+			if (lclu == first_lclu)
+				mapped = 1;
+		}
+	}
+
+out:
+	ext4_ext_drop_refs(path);
+	kfree(path);
+
+	return err ? err : mapped;
+}
diff --git a/fs/ext4/extents_status.c b/fs/ext4/extents_status.c
index 194785ce890a..c5d456e12062 100644
--- a/fs/ext4/extents_status.c
+++ b/fs/ext4/extents_status.c
@@ -1552,3 +1552,56 @@ bool ext4_is_pending(struct inode *inode, ext4_lblk_t lblk)
 
 	return ret;
 }
+
+/*
+ * ext4_es_insert_delayed_block - adds a delayed block to the extents status
+ *                                tree, adding a pending reservation where
+ *                                needed
+ *
+ * @inode - file containing the newly added block
+ * @lblk - logical block to be added
+ * @allocated - indicates whether a physical cluster has been allocated for
+ *              the logical cluster that contains the block
+ *
+ * Returns 0 on success, negative error code on failure.
+ */
+int ext4_es_insert_delayed_block(struct inode *inode, ext4_lblk_t lblk,
+				 bool allocated)
+{
+	struct extent_status newes;
+	int err = 0;
+
+	es_debug("add [%u/1) delayed to extent status tree of inode %lu\n",
+		 lblk, inode->i_ino);
+
+	newes.es_lblk = lblk;
+	newes.es_len = 1;
+	ext4_es_store_pblock_status(&newes, ~0, EXTENT_STATUS_DELAYED);
+	trace_ext4_es_insert_delayed_block(inode, &newes, allocated);
+
+	ext4_es_insert_extent_check(inode, &newes);
+
+	write_lock(&EXT4_I(inode)->i_es_lock);
+
+	err = __es_remove_extent(inode, lblk, lblk);
+	if (err != 0)
+		goto error;
+retry:
+	err = __es_insert_extent(inode, &newes);
+	if (err == -ENOMEM && __es_shrink(EXT4_SB(inode->i_sb),
+					  128, EXT4_I(inode)))
+		goto retry;
+	if (err != 0)
+		goto error;
+
+	if (allocated)
+		__insert_pending(inode, lblk);
+
+error:
+	write_unlock(&EXT4_I(inode)->i_es_lock);
+
+	ext4_es_print_tree(inode);
+	ext4_print_pending_tree(inode);
+
+	return err;
+}
diff --git a/fs/ext4/extents_status.h b/fs/ext4/extents_status.h
index 379b7171c67c..9d3c676ec623 100644
--- a/fs/ext4/extents_status.h
+++ b/fs/ext4/extents_status.h
@@ -178,6 +178,16 @@ static inline int ext4_es_is_hole(struct extent_status *es)
 	return (ext4_es_type(es) & EXTENT_STATUS_HOLE) != 0;
 }
 
+static inline int ext4_es_is_mapped(struct extent_status *es)
+{
+	return (ext4_es_is_written(es) || ext4_es_is_unwritten(es));
+}
+
+static inline int ext4_es_is_delonly(struct extent_status *es)
+{
+	return (ext4_es_is_delayed(es) && !ext4_es_is_unwritten(es));
+}
+
 static inline void ext4_es_set_referenced(struct extent_status *es)
 {
 	es->es_pblk |= ((ext4_fsblk_t)EXTENT_STATUS_REFERENCED) << ES_SHIFT;
@@ -232,5 +242,7 @@ extern void ext4_exit_pending(void);
 extern void ext4_init_pending_tree(struct ext4_pending_tree *tree);
 extern void ext4_remove_pending(struct inode *inode, ext4_lblk_t lblk);
 extern bool ext4_is_pending(struct inode *inode, ext4_lblk_t lblk);
+extern int ext4_es_insert_delayed_block(struct inode *inode, ext4_lblk_t lblk,
+					bool allocated);
 
 #endif /* _EXT4_EXTENTS_STATUS_H */
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index b83bf3308b5e..57c6dd38f071 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1781,6 +1781,65 @@ static int ext4_bh_delay_or_unwritten(handle_t *handle, struct buffer_head *bh)
 }
 
 /*
+ * ext4_insert_delayed_block - adds a delayed block to the extents status
+ *                             tree, incrementing the reserved cluster/block
+ *                             count or making a pending reservation
+ *                             where needed
+ *
+ * @inode - file containing the newly added block
+ * @lblk - logical block to be added
+ *
+ * Returns 0 on success, negative error code on failure.
+ */
+static int ext4_insert_delayed_block(struct inode *inode, ext4_lblk_t lblk)
+{
+	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
+	int ret;
+	bool allocated = false;
+
+	/*
+	 * If the cluster containing lblk is shared with a delayed,
+	 * written, or unwritten extent in a bigalloc file system, it's
+	 * already been accounted for and does not need to be reserved.
+	 * A pending reservation must be made for the cluster if it's
+	 * shared with a written or unwritten extent and doesn't already
+	 * have one.  Written and unwritten extents can be purged from the
+	 * extents status tree if the system is under memory pressure, so
+	 * it's necessary to examine the extent tree if a search of the
+	 * extents status tree doesn't get a match.
+	 */
+	if (sbi->s_cluster_ratio == 1) {
+		ret = ext4_da_reserve_space(inode);
+		if (ret != 0)   /* ENOSPC */
+			goto errout;
+	} else {   /* bigalloc */
+		if (!ext4_es_scan_clu(inode, &ext4_es_is_delonly, lblk)) {
+			if (!ext4_es_scan_clu(inode,
+					      &ext4_es_is_mapped, lblk)) {
+				ret = ext4_clu_mapped(inode,
+						      EXT4_B2C(sbi, lblk));
+				if (ret < 0)
+					goto errout;
+				if (ret == 0) {
+					ret = ext4_da_reserve_space(inode);
+					if (ret != 0)   /* ENOSPC */
+						goto errout;
+				} else {
+					allocated = true;
+				}
+			} else {
+				allocated = true;
+			}
+		}
+	}
+
+	ret = ext4_es_insert_delayed_block(inode, lblk, allocated);
+
+errout:
+	return ret;
+}
+
+/*
  * This function is grabs code from the very beginning of
  * ext4_map_blocks, but assumes that the caller is from delayed write
  * time. This function looks up the requested blocks and sets the
@@ -1864,25 +1923,9 @@ add_delayed:
 		 * XXX: __block_prepare_write() unmaps passed block,
 		 * is it OK?
 		 */
-		/*
-		 * If the block was allocated from previously allocated cluster,
-		 * then we don't need to reserve it again. However we still need
-		 * to reserve metadata for every block we're going to write.
-		 */
-		if (EXT4_SB(inode->i_sb)->s_cluster_ratio == 1 ||
-		    !ext4_es_scan_clu(inode,
-				      &ext4_es_is_delayed, map->m_lblk)) {
-			ret = ext4_da_reserve_space(inode);
-			if (ret) {
-				/* not enough space to reserve */
-				retval = ret;
-				goto out_unlock;
-			}
-		}
 
-		ret = ext4_es_insert_extent(inode, map->m_lblk, map->m_len,
-					    ~0, EXTENT_STATUS_DELAYED);
-		if (ret) {
+		ret = ext4_insert_delayed_block(inode, map->m_lblk);
+		if (ret != 0) {
 			retval = ret;
 			goto out_unlock;
 		}