1 files changed, 47 insertions, 18 deletions

diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 471d863958bc..36855c1f8daf 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -331,17 +331,34 @@ struct inode_switch_wbs_context {
 	struct work_struct	work;
 };
 
+static void bdi_down_write_wb_switch_rwsem(struct backing_dev_info *bdi)
+{
+	down_write(&bdi->wb_switch_rwsem);
+}
+
+static void bdi_up_write_wb_switch_rwsem(struct backing_dev_info *bdi)
+{
+	up_write(&bdi->wb_switch_rwsem);
+}
+
 static void inode_switch_wbs_work_fn(struct work_struct *work)
 {
 	struct inode_switch_wbs_context *isw =
 		container_of(work, struct inode_switch_wbs_context, work);
 	struct inode *inode = isw->inode;
+	struct backing_dev_info *bdi = inode_to_bdi(inode);
 	struct address_space *mapping = inode->i_mapping;
 	struct bdi_writeback *old_wb = inode->i_wb;
 	struct bdi_writeback *new_wb = isw->new_wb;
-	struct radix_tree_iter iter;
+	XA_STATE(xas, &mapping->i_pages, 0);
+	struct page *page;
 	bool switched = false;
-	void **slot;
+
+	/*
+	 * If @inode switches cgwb membership while sync_inodes_sb() is
+	 * being issued, sync_inodes_sb() might miss it.  Synchronize.
+	 */
+	down_read(&bdi->wb_switch_rwsem);
 
 	/*
 	 * By the time control reaches here, RCU grace period has passed
@@ -375,25 +392,18 @@ static void inode_switch_wbs_work_fn(struct work_struct *work)
 	 * to possibly dirty pages while PAGECACHE_TAG_WRITEBACK points to
 	 * pages actually under writeback.
 	 */
-	radix_tree_for_each_tagged(slot, &mapping->i_pages, &iter, 0,
-				   PAGECACHE_TAG_DIRTY) {
-		struct page *page = radix_tree_deref_slot_protected(slot,
-						&mapping->i_pages.xa_lock);
-		if (likely(page) && PageDirty(page)) {
+	xas_for_each_marked(&xas, page, ULONG_MAX, PAGECACHE_TAG_DIRTY) {
+		if (PageDirty(page)) {
 			dec_wb_stat(old_wb, WB_RECLAIMABLE);
 			inc_wb_stat(new_wb, WB_RECLAIMABLE);
 		}
 	}
 
-	radix_tree_for_each_tagged(slot, &mapping->i_pages, &iter, 0,
-				   PAGECACHE_TAG_WRITEBACK) {
-		struct page *page = radix_tree_deref_slot_protected(slot,
-						&mapping->i_pages.xa_lock);
-		if (likely(page)) {
-			WARN_ON_ONCE(!PageWriteback(page));
-			dec_wb_stat(old_wb, WB_WRITEBACK);
-			inc_wb_stat(new_wb, WB_WRITEBACK);
-		}
+	xas_set(&xas, 0);
+	xas_for_each_marked(&xas, page, ULONG_MAX, PAGECACHE_TAG_WRITEBACK) {
+		WARN_ON_ONCE(!PageWriteback(page));
+		dec_wb_stat(old_wb, WB_WRITEBACK);
+		inc_wb_stat(new_wb, WB_WRITEBACK);
 	}
 
 	wb_get(new_wb);
@@ -435,6 +445,8 @@ skip_switch:
 	spin_unlock(&new_wb->list_lock);
 	spin_unlock(&old_wb->list_lock);
 
+	up_read(&bdi->wb_switch_rwsem);
+
 	if (switched) {
 		wb_wakeup(new_wb);
 		wb_put(old_wb);
@@ -475,9 +487,18 @@ static void inode_switch_wbs(struct inode *inode, int new_wb_id)
 	if (inode->i_state & I_WB_SWITCH)
 		return;
 
+	/*
+	 * Avoid starting new switches while sync_inodes_sb() is in
+	 * progress.  Otherwise, if the down_write protected issue path
+	 * blocks heavily, we might end up starting a large number of
+	 * switches which will block on the rwsem.
+	 */
+	if (!down_read_trylock(&bdi->wb_switch_rwsem))
+		return;
+
 	isw = kzalloc(sizeof(*isw), GFP_ATOMIC);
 	if (!isw)
-		return;
+		goto out_unlock;
 
 	/* find and pin the new wb */
 	rcu_read_lock();
@@ -511,12 +532,14 @@ static void inode_switch_wbs(struct inode *inode, int new_wb_id)
 	 * Let's continue after I_WB_SWITCH is guaranteed to be visible.
 	 */
 	call_rcu(&isw->rcu_head, inode_switch_wbs_rcu_fn);
-	return;
+	goto out_unlock;
 
 out_free:
 	if (isw->new_wb)
 		wb_put(isw->new_wb);
 	kfree(isw);
+out_unlock:
+	up_read(&bdi->wb_switch_rwsem);
 }
 
 /**
@@ -894,6 +917,9 @@ fs_initcall(cgroup_writeback_init);
 
 #else	/* CONFIG_CGROUP_WRITEBACK */
 
+static void bdi_down_write_wb_switch_rwsem(struct backing_dev_info *bdi) { }
+static void bdi_up_write_wb_switch_rwsem(struct backing_dev_info *bdi) { }
+
 static struct bdi_writeback *
 locked_inode_to_wb_and_lock_list(struct inode *inode)
 	__releases(&inode->i_lock)
@@ -2420,8 +2446,11 @@ void sync_inodes_sb(struct super_block *sb)
 		return;
 	WARN_ON(!rwsem_is_locked(&sb->s_umount));
 
+	/* protect against inode wb switch, see inode_switch_wbs_work_fn() */
+	bdi_down_write_wb_switch_rwsem(bdi);
 	bdi_split_work_to_wbs(bdi, &work, false);
 	wb_wait_for_completion(bdi, &done);
+	bdi_up_write_wb_switch_rwsem(bdi);
 
 	wait_sb_inodes(sb);
 }