1 files changed, 106 insertions, 78 deletions

diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 30eceaf829a7..5df02c707aee 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -153,25 +153,30 @@ static struct btrfs_ordered_extent *alloc_ordered_extent(
 	struct btrfs_ordered_extent *entry;
 	int ret;
 	u64 qgroup_rsv = 0;
+	const bool is_nocow = (flags &
+	       ((1U << BTRFS_ORDERED_NOCOW) | (1U << BTRFS_ORDERED_PREALLOC)));
 
-	if (flags &
-	    ((1 << BTRFS_ORDERED_NOCOW) | (1 << BTRFS_ORDERED_PREALLOC))) {
-		/* For nocow write, we can release the qgroup rsv right now */
+	/*
+	 * For a NOCOW write we can free the qgroup reserve right now. For a COW
+	 * one we transfer the reserved space from the inode's iotree into the
+	 * ordered extent by calling btrfs_qgroup_release_data() and tracking
+	 * the qgroup reserved amount in the ordered extent, so that later after
+	 * completing the ordered extent, when running the data delayed ref it
+	 * creates, we free the reserved data with btrfs_qgroup_free_refroot().
+	 */
+	if (is_nocow)
 		ret = btrfs_qgroup_free_data(inode, NULL, file_offset, num_bytes, &qgroup_rsv);
-		if (ret < 0)
-			return ERR_PTR(ret);
-	} else {
-		/*
-		 * The ordered extent has reserved qgroup space, release now
-		 * and pass the reserved number for qgroup_record to free.
-		 */
+	else
 		ret = btrfs_qgroup_release_data(inode, file_offset, num_bytes, &qgroup_rsv);
-		if (ret < 0)
-			return ERR_PTR(ret);
-	}
+
+	if (ret < 0)
+		return ERR_PTR(ret);
+
 	entry = kmem_cache_zalloc(btrfs_ordered_extent_cache, GFP_NOFS);
-	if (!entry)
-		return ERR_PTR(-ENOMEM);
+	if (!entry) {
+		entry = ERR_PTR(-ENOMEM);
+		goto out;
+	}
 
 	entry->file_offset = file_offset;
 	entry->num_bytes = num_bytes;
@@ -180,7 +185,12 @@ static struct btrfs_ordered_extent *alloc_ordered_extent(
 	entry->disk_num_bytes = disk_num_bytes;
 	entry->offset = offset;
 	entry->bytes_left = num_bytes;
-	entry->inode = BTRFS_I(igrab(&inode->vfs_inode));
+	if (WARN_ON_ONCE(!igrab(&inode->vfs_inode))) {
+		kmem_cache_free(btrfs_ordered_extent_cache, entry);
+		entry = ERR_PTR(-ESTALE);
+		goto out;
+	}
+	entry->inode = inode;
 	entry->compress_type = compress_type;
 	entry->truncated_len = (u64)-1;
 	entry->qgroup_rsv = qgroup_rsv;
@@ -203,6 +213,12 @@ static struct btrfs_ordered_extent *alloc_ordered_extent(
 	btrfs_mod_outstanding_extents(inode, 1);
 	spin_unlock(&inode->lock);
 
+out:
+	if (IS_ERR(entry) && !is_nocow)
+		btrfs_qgroup_free_refroot(inode->root->fs_info,
+					  btrfs_root_id(inode->root),
+					  qgroup_rsv, BTRFS_QGROUP_RSV_DATA);
+
 	return entry;
 }
 
@@ -221,14 +237,14 @@ static void insert_ordered_extent(struct btrfs_ordered_extent *entry)
 	/* One ref for the tree. */
 	refcount_inc(&entry->refs);
 
-	spin_lock_irq(&inode->ordered_tree_lock);
+	spin_lock(&inode->ordered_tree_lock);
 	node = tree_insert(&inode->ordered_tree, entry->file_offset,
 			   &entry->rb_node);
 	if (unlikely(node))
 		btrfs_panic(fs_info, -EEXIST,
 				"inconsistency in ordered tree at offset %llu",
 				entry->file_offset);
-	spin_unlock_irq(&inode->ordered_tree_lock);
+	spin_unlock(&inode->ordered_tree_lock);
 
 	spin_lock(&root->ordered_extent_lock);
 	list_add_tail(&entry->root_extent_list,
@@ -253,7 +269,7 @@ static void insert_ordered_extent(struct btrfs_ordered_extent *entry)
  * @disk_bytenr:     Offset of extent on disk.
  * @disk_num_bytes:  Size of extent on disk.
  * @offset:          Offset into unencoded data where file data starts.
- * @flags:           Flags specifying type of extent (1 << BTRFS_ORDERED_*).
+ * @flags:           Flags specifying type of extent (1U << BTRFS_ORDERED_*).
  * @compress_type:   Compression algorithm used for data.
  *
  * Most of these parameters correspond to &struct btrfs_file_extent_item. The
@@ -312,9 +328,9 @@ void btrfs_add_ordered_sum(struct btrfs_ordered_extent *entry,
 {
 	struct btrfs_inode *inode = entry->inode;
 
-	spin_lock_irq(&inode->ordered_tree_lock);
+	spin_lock(&inode->ordered_tree_lock);
 	list_add_tail(&sum->list, &entry->list);
-	spin_unlock_irq(&inode->ordered_tree_lock);
+	spin_unlock(&inode->ordered_tree_lock);
 }
 
 void btrfs_mark_ordered_extent_error(struct btrfs_ordered_extent *ordered)
@@ -343,7 +359,7 @@ static bool can_finish_ordered_extent(struct btrfs_ordered_extent *ordered,
 	if (folio) {
 		ASSERT(folio->mapping);
 		ASSERT(folio_pos(folio) <= file_offset);
-		ASSERT(file_offset + len <= folio_pos(folio) + folio_size(folio));
+		ASSERT(file_offset + len <= folio_next_pos(folio));
 
 		/*
 		 * Ordered flag indicates whether we still have
@@ -401,15 +417,14 @@ void btrfs_finish_ordered_extent(struct btrfs_ordered_extent *ordered,
 				 bool uptodate)
 {
 	struct btrfs_inode *inode = ordered->inode;
-	unsigned long flags;
 	bool ret;
 
 	trace_btrfs_finish_ordered_extent(inode, file_offset, len, uptodate);
 
-	spin_lock_irqsave(&inode->ordered_tree_lock, flags);
+	spin_lock(&inode->ordered_tree_lock);
 	ret = can_finish_ordered_extent(ordered, folio, file_offset, len,
 					uptodate);
-	spin_unlock_irqrestore(&inode->ordered_tree_lock, flags);
+	spin_unlock(&inode->ordered_tree_lock);
 
 	/*
 	 * If this is a COW write it means we created new extent maps for the
@@ -465,18 +480,16 @@ void btrfs_mark_ordered_io_finished(struct btrfs_inode *inode,
 {
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
-	unsigned long flags;
 	u64 cur = file_offset;
+	const u64 end = file_offset + num_bytes;
 
-	trace_btrfs_writepage_end_io_hook(inode, file_offset,
-					  file_offset + num_bytes - 1,
-					  uptodate);
+	trace_btrfs_writepage_end_io_hook(inode, file_offset, end - 1, uptodate);
 
-	spin_lock_irqsave(&inode->ordered_tree_lock, flags);
-	while (cur < file_offset + num_bytes) {
+	spin_lock(&inode->ordered_tree_lock);
+	while (cur < end) {
 		u64 entry_end;
-		u64 end;
-		u32 len;
+		u64 this_end;
+		u64 len;
 
 		node = ordered_tree_search(inode, cur);
 		/* No ordered extents at all */
@@ -519,19 +532,18 @@ void btrfs_mark_ordered_io_finished(struct btrfs_inode *inode,
 		 *	|
 		 *	cur
 		 */
-		end = min(entry->file_offset + entry->num_bytes,
-			  file_offset + num_bytes) - 1;
-		ASSERT(end + 1 - cur < U32_MAX);
-		len = end + 1 - cur;
+		this_end = min(entry_end, end);
+		len = this_end - cur;
+		ASSERT(len < U32_MAX);
 
 		if (can_finish_ordered_extent(entry, folio, cur, len, uptodate)) {
-			spin_unlock_irqrestore(&inode->ordered_tree_lock, flags);
+			spin_unlock(&inode->ordered_tree_lock);
 			btrfs_queue_ordered_fn(entry);
-			spin_lock_irqsave(&inode->ordered_tree_lock, flags);
+			spin_lock(&inode->ordered_tree_lock);
 		}
 		cur += len;
 	}
-	spin_unlock_irqrestore(&inode->ordered_tree_lock, flags);
+	spin_unlock(&inode->ordered_tree_lock);
 }
 
 /*
@@ -557,10 +569,9 @@ bool btrfs_dec_test_ordered_pending(struct btrfs_inode *inode,
 {
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
-	unsigned long flags;
 	bool finished = false;
 
-	spin_lock_irqsave(&inode->ordered_tree_lock, flags);
+	spin_lock(&inode->ordered_tree_lock);
 	if (cached && *cached) {
 		entry = *cached;
 		goto have_entry;
@@ -597,7 +608,7 @@ out:
 		refcount_inc(&entry->refs);
 		trace_btrfs_ordered_extent_dec_test_pending(inode, entry);
 	}
-	spin_unlock_irqrestore(&inode->ordered_tree_lock, flags);
+	spin_unlock(&inode->ordered_tree_lock);
 	return finished;
 }
 
@@ -607,23 +618,18 @@ out:
  */
 void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 {
-	struct list_head *cur;
-	struct btrfs_ordered_sum *sum;
-
 	trace_btrfs_ordered_extent_put(entry->inode, entry);
 
 	if (refcount_dec_and_test(&entry->refs)) {
+		struct btrfs_ordered_sum *sum;
+		struct btrfs_ordered_sum *tmp;
+
 		ASSERT(list_empty(&entry->root_extent_list));
 		ASSERT(list_empty(&entry->log_list));
 		ASSERT(RB_EMPTY_NODE(&entry->rb_node));
-		if (entry->inode)
-			btrfs_add_delayed_iput(entry->inode);
-		while (!list_empty(&entry->list)) {
-			cur = entry->list.next;
-			sum = list_entry(cur, struct btrfs_ordered_sum, list);
-			list_del(&sum->list);
+		btrfs_add_delayed_iput(entry->inode);
+		list_for_each_entry_safe(sum, tmp, &entry->list, list)
 			kvfree(sum);
-		}
 		kmem_cache_free(btrfs_ordered_extent_cache, entry);
 	}
 }
@@ -667,7 +673,7 @@ void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
 	percpu_counter_add_batch(&fs_info->ordered_bytes, -entry->num_bytes,
 				 fs_info->delalloc_batch);
 
-	spin_lock_irq(&btrfs_inode->ordered_tree_lock);
+	spin_lock(&btrfs_inode->ordered_tree_lock);
 	node = &entry->rb_node;
 	rb_erase(node, &btrfs_inode->ordered_tree);
 	RB_CLEAR_NODE(node);
@@ -675,7 +681,7 @@ void btrfs_remove_ordered_extent(struct btrfs_inode *btrfs_inode,
 		btrfs_inode->ordered_tree_last = NULL;
 	set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
 	pending = test_and_clear_bit(BTRFS_ORDERED_PENDING, &entry->flags);
-	spin_unlock_irq(&btrfs_inode->ordered_tree_lock);
+	spin_unlock(&btrfs_inode->ordered_tree_lock);
 
 	/*
 	 * The current running transaction is waiting on us, we need to let it
@@ -842,10 +848,12 @@ void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, u64 nr,
 /*
  * Start IO and wait for a given ordered extent to finish.
  *
- * Wait on page writeback for all the pages in the extent and the IO completion
- * code to insert metadata into the btree corresponding to the extent.
+ * Wait on page writeback for all the pages in the extent but not in
+ * [@nowriteback_start, @nowriteback_start + @nowriteback_len) and the
+ * IO completion code to insert metadata into the btree corresponding to the extent.
  */
-void btrfs_start_ordered_extent(struct btrfs_ordered_extent *entry)
+void btrfs_start_ordered_extent_nowriteback(struct btrfs_ordered_extent *entry,
+					    u64 nowriteback_start, u32 nowriteback_len)
 {
 	u64 start = entry->file_offset;
 	u64 end = start + entry->num_bytes - 1;
@@ -865,8 +873,19 @@ void btrfs_start_ordered_extent(struct btrfs_ordered_extent *entry)
 	 * start IO on any dirty ones so the wait doesn't stall waiting
 	 * for the flusher thread to find them
 	 */
-	if (!test_bit(BTRFS_ORDERED_DIRECT, &entry->flags))
-		filemap_fdatawrite_range(inode->vfs_inode.i_mapping, start, end);
+	if (!test_bit(BTRFS_ORDERED_DIRECT, &entry->flags)) {
+		if (!nowriteback_len) {
+			filemap_fdatawrite_range(inode->vfs_inode.i_mapping, start, end);
+		} else {
+			if (start < nowriteback_start)
+				filemap_fdatawrite_range(inode->vfs_inode.i_mapping, start,
+							 nowriteback_start - 1);
+			if (nowriteback_start + nowriteback_len < end)
+				filemap_fdatawrite_range(inode->vfs_inode.i_mapping,
+							 nowriteback_start + nowriteback_len,
+							 end);
+		}
+	}
 
 	if (!freespace_inode)
 		btrfs_might_wait_for_event(inode->root->fs_info, btrfs_ordered_extent);
@@ -947,9 +966,8 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct btrfs_inode *ino
 {
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
-	unsigned long flags;
 
-	spin_lock_irqsave(&inode->ordered_tree_lock, flags);
+	spin_lock(&inode->ordered_tree_lock);
 	node = ordered_tree_search(inode, file_offset);
 	if (!node)
 		goto out;
@@ -962,7 +980,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct btrfs_inode *ino
 		trace_btrfs_ordered_extent_lookup(inode, entry);
 	}
 out:
-	spin_unlock_irqrestore(&inode->ordered_tree_lock, flags);
+	spin_unlock(&inode->ordered_tree_lock);
 	return entry;
 }
 
@@ -975,7 +993,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_range(
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
 
-	spin_lock_irq(&inode->ordered_tree_lock);
+	spin_lock(&inode->ordered_tree_lock);
 	node = ordered_tree_search(inode, file_offset);
 	if (!node) {
 		node = ordered_tree_search(inode, file_offset + len);
@@ -1002,7 +1020,7 @@ out:
 		refcount_inc(&entry->refs);
 		trace_btrfs_ordered_extent_lookup_range(inode, entry);
 	}
-	spin_unlock_irq(&inode->ordered_tree_lock);
+	spin_unlock(&inode->ordered_tree_lock);
 	return entry;
 }
 
@@ -1017,7 +1035,7 @@ void btrfs_get_ordered_extents_for_logging(struct btrfs_inode *inode,
 
 	btrfs_assert_inode_locked(inode);
 
-	spin_lock_irq(&inode->ordered_tree_lock);
+	spin_lock(&inode->ordered_tree_lock);
 	for (n = rb_first(&inode->ordered_tree); n; n = rb_next(n)) {
 		struct btrfs_ordered_extent *ordered;
 
@@ -1031,7 +1049,7 @@ void btrfs_get_ordered_extents_for_logging(struct btrfs_inode *inode,
 		refcount_inc(&ordered->refs);
 		trace_btrfs_ordered_extent_lookup_for_logging(inode, ordered);
 	}
-	spin_unlock_irq(&inode->ordered_tree_lock);
+	spin_unlock(&inode->ordered_tree_lock);
 }
 
 /*
@@ -1044,7 +1062,7 @@ btrfs_lookup_first_ordered_extent(struct btrfs_inode *inode, u64 file_offset)
 	struct rb_node *node;
 	struct btrfs_ordered_extent *entry = NULL;
 
-	spin_lock_irq(&inode->ordered_tree_lock);
+	spin_lock(&inode->ordered_tree_lock);
 	node = ordered_tree_search(inode, file_offset);
 	if (!node)
 		goto out;
@@ -1053,7 +1071,7 @@ btrfs_lookup_first_ordered_extent(struct btrfs_inode *inode, u64 file_offset)
 	refcount_inc(&entry->refs);
 	trace_btrfs_ordered_extent_lookup_first(inode, entry);
 out:
-	spin_unlock_irq(&inode->ordered_tree_lock);
+	spin_unlock(&inode->ordered_tree_lock);
 	return entry;
 }
 
@@ -1075,7 +1093,7 @@ struct btrfs_ordered_extent *btrfs_lookup_first_ordered_range(
 	struct rb_node *next;
 	struct btrfs_ordered_extent *entry = NULL;
 
-	spin_lock_irq(&inode->ordered_tree_lock);
+	spin_lock(&inode->ordered_tree_lock);
 	node = inode->ordered_tree.rb_node;
 	/*
 	 * Here we don't want to use tree_search() which will use tree->last
@@ -1130,7 +1148,7 @@ out:
 		trace_btrfs_ordered_extent_lookup_first_range(inode, entry);
 	}
 
-	spin_unlock_irq(&inode->ordered_tree_lock);
+	spin_unlock(&inode->ordered_tree_lock);
 	return entry;
 }
 
@@ -1160,7 +1178,7 @@ void btrfs_lock_and_flush_ordered_range(struct btrfs_inode *inode, u64 start,
 		cachedp = cached_state;
 
 	while (1) {
-		lock_extent(&inode->io_tree, start, end, cachedp);
+		btrfs_lock_extent(&inode->io_tree, start, end, cachedp);
 		ordered = btrfs_lookup_ordered_range(inode, start,
 						     end - start + 1);
 		if (!ordered) {
@@ -1173,7 +1191,7 @@ void btrfs_lock_and_flush_ordered_range(struct btrfs_inode *inode, u64 start,
 				refcount_dec(&cache->refs);
 			break;
 		}
-		unlock_extent(&inode->io_tree, start, end, cachedp);
+		btrfs_unlock_extent(&inode->io_tree, start, end, cachedp);
 		btrfs_start_ordered_extent(ordered);
 		btrfs_put_ordered_extent(ordered);
 	}
@@ -1191,7 +1209,7 @@ bool btrfs_try_lock_ordered_range(struct btrfs_inode *inode, u64 start, u64 end,
 {
 	struct btrfs_ordered_extent *ordered;
 
-	if (!try_lock_extent(&inode->io_tree, start, end, cached_state))
+	if (!btrfs_try_lock_extent(&inode->io_tree, start, end, cached_state))
 		return false;
 
 	ordered = btrfs_lookup_ordered_range(inode, start, end - start + 1);
@@ -1199,7 +1217,7 @@ bool btrfs_try_lock_ordered_range(struct btrfs_inode *inode, u64 start, u64 end,
 		return true;
 
 	btrfs_put_ordered_extent(ordered);
-	unlock_extent(&inode->io_tree, start, end, cached_state);
+	btrfs_unlock_extent(&inode->io_tree, start, end, cached_state);
 
 	return false;
 }
@@ -1229,6 +1247,18 @@ struct btrfs_ordered_extent *btrfs_split_ordered_extent(
 	 */
 	if (WARN_ON_ONCE(len >= ordered->num_bytes))
 		return ERR_PTR(-EINVAL);
+	/*
+	 * If our ordered extent had an error there's no point in continuing.
+	 * The error may have come from a transaction abort done either by this
+	 * task or some other concurrent task, and the transaction abort path
+	 * iterates over all existing ordered extents and sets the flag
+	 * BTRFS_ORDERED_IOERR on them.
+	 */
+	if (unlikely(flags & (1U << BTRFS_ORDERED_IOERR))) {
+		const int fs_error = BTRFS_FS_ERROR(fs_info);
+
+		return fs_error ? ERR_PTR(fs_error) : ERR_PTR(-EIO);
+	}
 	/* We cannot split partially completed ordered extents. */
 	if (ordered->bytes_left) {
 		ASSERT(!(flags & ~BTRFS_ORDERED_TYPE_FLAGS));
@@ -1250,9 +1280,7 @@ struct btrfs_ordered_extent *btrfs_split_ordered_extent(
 	/*
 	 * Take the root's ordered_extent_lock to avoid a race with
 	 * btrfs_wait_ordered_extents() when updating the disk_bytenr and
-	 * disk_num_bytes fields of the ordered extent below. And we disable
-	 * IRQs because the inode's ordered_tree_lock is used in IRQ context
-	 * elsewhere.
+	 * disk_num_bytes fields of the ordered extent below.
 	 *
 	 * There's no concern about a previous caller of
 	 * btrfs_wait_ordered_extents() getting the trimmed ordered extent