Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs updates from Chris Mason:
 "The biggest change here is Josef's rework of the btrfs quota
  accounting, which improves the in-memory tracking of delayed extent
  operations.

  I had been working on Btrfs stack usage for a while, mostly because it
  had become impossible to do long stress runs with slab, lockdep and
  pagealloc debugging turned on without blowing the stack.  Even though
  you upgraded us to a nice king sized stack, I kept most of the
  patches.

  We also have some very hard to find corruption fixes, an awesome sysfs
  use after free, and the usual assortment of optimizations, cleanups
  and other fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (80 commits)
  Btrfs: convert smp_mb__{before,after}_clear_bit
  Btrfs: fix scrub_print_warning to handle skinny metadata extents
  Btrfs: make fsync work after cloning into a file
  Btrfs: use right type to get real comparison
  Btrfs: don't check nodes for extent items
  Btrfs: don't release invalid page in btrfs_page_exists_in_range()
  Btrfs: make sure we retry if page is a retriable exception
  Btrfs: make sure we retry if we couldn't get the page
  btrfs: replace EINVAL with EOPNOTSUPP for dev_replace raid56
  trivial: fs/btrfs/ioctl.c: fix typo s/substract/subtract/
  Btrfs: fix leaf corruption after __btrfs_drop_extents
  Btrfs: ensure btrfs_prev_leaf doesn't miss 1 item
  Btrfs: fix clone to deal with holes when NO_HOLES feature is enabled
  btrfs: free delayed node outside of root->inode_lock
  btrfs: replace EINVAL with ERANGE for resize when ULLONG_MAX
  Btrfs: fix transaction leak during fsync call
  btrfs: Avoid trucating page or punching hole in a already existed hole.
  Btrfs: update commit root on snapshot creation after orphan cleanup
  Btrfs: ioctl, don't re-lock extent range when not necessary
  Btrfs: avoid visiting all extent items when cloning a range
  ...

1 files changed, 257 insertions, 144 deletions

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 4cd0ac983f91..f25a9092b946 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -1693,6 +1693,7 @@ again:
 		 * shortening the size of the delalloc range we're searching
 		 */
 		free_extent_state(cached_state);
+		cached_state = NULL;
 		if (!loops) {
 			max_bytes = PAGE_CACHE_SIZE;
 			loops = 1;
@@ -2367,6 +2368,8 @@ int end_extent_writepage(struct page *page, int err, u64 start, u64 end)
 	if (!uptodate) {
 		ClearPageUptodate(page);
 		SetPageError(page);
+		ret = ret < 0 ? ret : -EIO;
+		mapping_set_error(page->mapping, ret);
 	}
 	return 0;
 }
@@ -3098,143 +3101,130 @@ static noinline void update_nr_written(struct page *page,
 }
 
 /*
- * the writepage semantics are similar to regular writepage.  extent
- * records are inserted to lock ranges in the tree, and as dirty areas
- * are found, they are marked writeback.  Then the lock bits are removed
- * and the end_io handler clears the writeback ranges
+ * helper for __extent_writepage, doing all of the delayed allocation setup.
+ *
+ * This returns 1 if our fill_delalloc function did all the work required
+ * to write the page (copy into inline extent).  In this case the IO has
+ * been started and the page is already unlocked.
+ *
+ * This returns 0 if all went well (page still locked)
+ * This returns < 0 if there were errors (page still locked)
  */
-static int __extent_writepage(struct page *page, struct writeback_control *wbc,
-			      void *data)
+static noinline_for_stack int writepage_delalloc(struct inode *inode,
+			      struct page *page, struct writeback_control *wbc,
+			      struct extent_page_data *epd,
+			      u64 delalloc_start,
+			      unsigned long *nr_written)
+{
+	struct extent_io_tree *tree = epd->tree;
+	u64 page_end = delalloc_start + PAGE_CACHE_SIZE - 1;
+	u64 nr_delalloc;
+	u64 delalloc_to_write = 0;
+	u64 delalloc_end = 0;
+	int ret;
+	int page_started = 0;
+
+	if (epd->extent_locked || !tree->ops || !tree->ops->fill_delalloc)
+		return 0;
+
+	while (delalloc_end < page_end) {
+		nr_delalloc = find_lock_delalloc_range(inode, tree,
+					       page,
+					       &delalloc_start,
+					       &delalloc_end,
+					       128 * 1024 * 1024);
+		if (nr_delalloc == 0) {
+			delalloc_start = delalloc_end + 1;
+			continue;
+		}
+		ret = tree->ops->fill_delalloc(inode, page,
+					       delalloc_start,
+					       delalloc_end,
+					       &page_started,
+					       nr_written);
+		/* File system has been set read-only */
+		if (ret) {
+			SetPageError(page);
+			/* fill_delalloc should be return < 0 for error
+			 * but just in case, we use > 0 here meaning the
+			 * IO is started, so we don't want to return > 0
+			 * unless things are going well.
+			 */
+			ret = ret < 0 ? ret : -EIO;
+			goto done;
+		}
+		/*
+		 * delalloc_end is already one less than the total
+		 * length, so we don't subtract one from
+		 * PAGE_CACHE_SIZE
+		 */
+		delalloc_to_write += (delalloc_end - delalloc_start +
+				      PAGE_CACHE_SIZE) >>
+				      PAGE_CACHE_SHIFT;
+		delalloc_start = delalloc_end + 1;
+	}
+	if (wbc->nr_to_write < delalloc_to_write) {
+		int thresh = 8192;
+
+		if (delalloc_to_write < thresh * 2)
+			thresh = delalloc_to_write;
+		wbc->nr_to_write = min_t(u64, delalloc_to_write,
+					 thresh);
+	}
+
+	/* did the fill delalloc function already unlock and start
+	 * the IO?
+	 */
+	if (page_started) {
+		/*
+		 * we've unlocked the page, so we can't update
+		 * the mapping's writeback index, just update
+		 * nr_to_write.
+		 */
+		wbc->nr_to_write -= *nr_written;
+		return 1;
+	}
+
+	ret = 0;
+
+done:
+	return ret;
+}
+
+/*
+ * helper for __extent_writepage.  This calls the writepage start hooks,
+ * and does the loop to map the page into extents and bios.
+ *
+ * We return 1 if the IO is started and the page is unlocked,
+ * 0 if all went well (page still locked)
+ * < 0 if there were errors (page still locked)
+ */
+static noinline_for_stack int __extent_writepage_io(struct inode *inode,
+				 struct page *page,
+				 struct writeback_control *wbc,
+				 struct extent_page_data *epd,
+				 loff_t i_size,
+				 unsigned long nr_written,
+				 int write_flags, int *nr_ret)
 {
-	struct inode *inode = page->mapping->host;
-	struct extent_page_data *epd = data;
 	struct extent_io_tree *tree = epd->tree;
 	u64 start = page_offset(page);
-	u64 delalloc_start;
 	u64 page_end = start + PAGE_CACHE_SIZE - 1;
 	u64 end;
 	u64 cur = start;
 	u64 extent_offset;
-	u64 last_byte = i_size_read(inode);
 	u64 block_start;
 	u64 iosize;
 	sector_t sector;
 	struct extent_state *cached_state = NULL;
 	struct extent_map *em;
 	struct block_device *bdev;
-	int ret;
-	int nr = 0;
 	size_t pg_offset = 0;
 	size_t blocksize;
-	loff_t i_size = i_size_read(inode);
-	unsigned long end_index = i_size >> PAGE_CACHE_SHIFT;
-	u64 nr_delalloc;
-	u64 delalloc_end;
-	int page_started;
-	int compressed;
-	int write_flags;
-	unsigned long nr_written = 0;
-	bool fill_delalloc = true;
-
-	if (wbc->sync_mode == WB_SYNC_ALL)
-		write_flags = WRITE_SYNC;
-	else
-		write_flags = WRITE;
-
-	trace___extent_writepage(page, inode, wbc);
-
-	WARN_ON(!PageLocked(page));
-
-	ClearPageError(page);
-
-	pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
-	if (page->index > end_index ||
-	   (page->index == end_index && !pg_offset)) {
-		page->mapping->a_ops->invalidatepage(page, 0, PAGE_CACHE_SIZE);
-		unlock_page(page);
-		return 0;
-	}
-
-	if (page->index == end_index) {
-		char *userpage;
-
-		userpage = kmap_atomic(page);
-		memset(userpage + pg_offset, 0,
-		       PAGE_CACHE_SIZE - pg_offset);
-		kunmap_atomic(userpage);
-		flush_dcache_page(page);
-	}
-	pg_offset = 0;
-
-	set_page_extent_mapped(page);
-
-	if (!tree->ops || !tree->ops->fill_delalloc)
-		fill_delalloc = false;
-
-	delalloc_start = start;
-	delalloc_end = 0;
-	page_started = 0;
-	if (!epd->extent_locked && fill_delalloc) {
-		u64 delalloc_to_write = 0;
-		/*
-		 * make sure the wbc mapping index is at least updated
-		 * to this page.
-		 */
-		update_nr_written(page, wbc, 0);
-
-		while (delalloc_end < page_end) {
-			nr_delalloc = find_lock_delalloc_range(inode, tree,
-						       page,
-						       &delalloc_start,
-						       &delalloc_end,
-						       128 * 1024 * 1024);
-			if (nr_delalloc == 0) {
-				delalloc_start = delalloc_end + 1;
-				continue;
-			}
-			ret = tree->ops->fill_delalloc(inode, page,
-						       delalloc_start,
-						       delalloc_end,
-						       &page_started,
-						       &nr_written);
-			/* File system has been set read-only */
-			if (ret) {
-				SetPageError(page);
-				goto done;
-			}
-			/*
-			 * delalloc_end is already one less than the total
-			 * length, so we don't subtract one from
-			 * PAGE_CACHE_SIZE
-			 */
-			delalloc_to_write += (delalloc_end - delalloc_start +
-					      PAGE_CACHE_SIZE) >>
-					      PAGE_CACHE_SHIFT;
-			delalloc_start = delalloc_end + 1;
-		}
-		if (wbc->nr_to_write < delalloc_to_write) {
-			int thresh = 8192;
-
-			if (delalloc_to_write < thresh * 2)
-				thresh = delalloc_to_write;
-			wbc->nr_to_write = min_t(u64, delalloc_to_write,
-						 thresh);
-		}
+	int ret = 0;
+	int nr = 0;
+	bool compressed;
 
-		/* did the fill delalloc function already unlock and start
-		 * the IO?
-		 */
-		if (page_started) {
-			ret = 0;
-			/*
-			 * we've unlocked the page, so we can't update
-			 * the mapping's writeback index, just update
-			 * nr_to_write.
-			 */
-			wbc->nr_to_write -= nr_written;
-			goto done_unlocked;
-		}
-	}
 	if (tree->ops && tree->ops->writepage_start_hook) {
 		ret = tree->ops->writepage_start_hook(page, start,
 						      page_end);
@@ -3244,9 +3234,10 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 				wbc->pages_skipped++;
 			else
 				redirty_page_for_writepage(wbc, page);
+
 			update_nr_written(page, wbc, nr_written);
 			unlock_page(page);
-			ret = 0;
+			ret = 1;
 			goto done_unlocked;
 		}
 	}
@@ -3258,7 +3249,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 	update_nr_written(page, wbc, nr_written + 1);
 
 	end = page_end;
-	if (last_byte <= start) {
+	if (i_size <= start) {
 		if (tree->ops && tree->ops->writepage_end_io_hook)
 			tree->ops->writepage_end_io_hook(page, start,
 							 page_end, NULL, 1);
@@ -3268,7 +3259,8 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 	blocksize = inode->i_sb->s_blocksize;
 
 	while (cur <= end) {
-		if (cur >= last_byte) {
+		u64 em_end;
+		if (cur >= i_size) {
 			if (tree->ops && tree->ops->writepage_end_io_hook)
 				tree->ops->writepage_end_io_hook(page, cur,
 							 page_end, NULL, 1);
@@ -3278,13 +3270,15 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 				     end - cur + 1, 1);
 		if (IS_ERR_OR_NULL(em)) {
 			SetPageError(page);
+			ret = PTR_ERR_OR_ZERO(em);
 			break;
 		}
 
 		extent_offset = cur - em->start;
-		BUG_ON(extent_map_end(em) <= cur);
+		em_end = extent_map_end(em);
+		BUG_ON(em_end <= cur);
 		BUG_ON(end < cur);
-		iosize = min(extent_map_end(em) - cur, end - cur + 1);
+		iosize = min(em_end - cur, end - cur + 1);
 		iosize = ALIGN(iosize, blocksize);
 		sector = (em->block_start + extent_offset) >> 9;
 		bdev = em->bdev;
@@ -3320,13 +3314,6 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 			pg_offset += iosize;
 			continue;
 		}
-		/* leave this out until we have a page_mkwrite call */
-		if (0 && !test_range_bit(tree, cur, cur + iosize - 1,
-				   EXTENT_DIRTY, 0, NULL)) {
-			cur = cur + iosize;
-			pg_offset += iosize;
-			continue;
-		}
 
 		if (tree->ops && tree->ops->writepage_io_hook) {
 			ret = tree->ops->writepage_io_hook(page, cur,
@@ -3337,7 +3324,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 		if (ret) {
 			SetPageError(page);
 		} else {
-			unsigned long max_nr = end_index + 1;
+			unsigned long max_nr = (i_size >> PAGE_CACHE_SHIFT) + 1;
 
 			set_range_writeback(tree, cur, cur + iosize - 1);
 			if (!PageWriteback(page)) {
@@ -3359,17 +3346,94 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 		nr++;
 	}
 done:
+	*nr_ret = nr;
+
+done_unlocked:
+
+	/* drop our reference on any cached states */
+	free_extent_state(cached_state);
+	return ret;
+}
+
+/*
+ * the writepage semantics are similar to regular writepage.  extent
+ * records are inserted to lock ranges in the tree, and as dirty areas
+ * are found, they are marked writeback.  Then the lock bits are removed
+ * and the end_io handler clears the writeback ranges
+ */
+static int __extent_writepage(struct page *page, struct writeback_control *wbc,
+			      void *data)
+{
+	struct inode *inode = page->mapping->host;
+	struct extent_page_data *epd = data;
+	u64 start = page_offset(page);
+	u64 page_end = start + PAGE_CACHE_SIZE - 1;
+	int ret;
+	int nr = 0;
+	size_t pg_offset = 0;
+	loff_t i_size = i_size_read(inode);
+	unsigned long end_index = i_size >> PAGE_CACHE_SHIFT;
+	int write_flags;
+	unsigned long nr_written = 0;
+
+	if (wbc->sync_mode == WB_SYNC_ALL)
+		write_flags = WRITE_SYNC;
+	else
+		write_flags = WRITE;
+
+	trace___extent_writepage(page, inode, wbc);
+
+	WARN_ON(!PageLocked(page));
+
+	ClearPageError(page);
+
+	pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
+	if (page->index > end_index ||
+	   (page->index == end_index && !pg_offset)) {
+		page->mapping->a_ops->invalidatepage(page, 0, PAGE_CACHE_SIZE);
+		unlock_page(page);
+		return 0;
+	}
+
+	if (page->index == end_index) {
+		char *userpage;
+
+		userpage = kmap_atomic(page);
+		memset(userpage + pg_offset, 0,
+		       PAGE_CACHE_SIZE - pg_offset);
+		kunmap_atomic(userpage);
+		flush_dcache_page(page);
+	}
+
+	pg_offset = 0;
+
+	set_page_extent_mapped(page);
+
+	ret = writepage_delalloc(inode, page, wbc, epd, start, &nr_written);
+	if (ret == 1)
+		goto done_unlocked;
+	if (ret)
+		goto done;
+
+	ret = __extent_writepage_io(inode, page, wbc, epd,
+				    i_size, nr_written, write_flags, &nr);
+	if (ret == 1)
+		goto done_unlocked;
+
+done:
 	if (nr == 0) {
 		/* make sure the mapping tag for page dirty gets cleared */
 		set_page_writeback(page);
 		end_page_writeback(page);
 	}
+	if (PageError(page)) {
+		ret = ret < 0 ? ret : -EIO;
+		end_extent_writepage(page, ret, start, page_end);
+	}
 	unlock_page(page);
+	return ret;
 
 done_unlocked:
-
-	/* drop our reference on any cached states */
-	free_extent_state(cached_state);
 	return 0;
 }
 
@@ -3385,9 +3449,10 @@ void wait_on_extent_buffer_writeback(struct extent_buffer *eb)
 		    TASK_UNINTERRUPTIBLE);
 }
 
-static int lock_extent_buffer_for_io(struct extent_buffer *eb,
-				     struct btrfs_fs_info *fs_info,
-				     struct extent_page_data *epd)
+static noinline_for_stack int
+lock_extent_buffer_for_io(struct extent_buffer *eb,
+			  struct btrfs_fs_info *fs_info,
+			  struct extent_page_data *epd)
 {
 	unsigned long i, num_pages;
 	int flush = 0;
@@ -3492,7 +3557,7 @@ static void end_bio_extent_buffer_writepage(struct bio *bio, int err)
 	bio_put(bio);
 }
 
-static int write_one_eb(struct extent_buffer *eb,
+static noinline_for_stack int write_one_eb(struct extent_buffer *eb,
 			struct btrfs_fs_info *fs_info,
 			struct writeback_control *wbc,
 			struct extent_page_data *epd)
@@ -3690,6 +3755,7 @@ static int extent_write_cache_pages(struct extent_io_tree *tree,
 	struct inode *inode = mapping->host;
 	int ret = 0;
 	int done = 0;
+	int err = 0;
 	int nr_to_write_done = 0;
 	struct pagevec pvec;
 	int nr_pages;
@@ -3776,8 +3842,8 @@ retry:
 				unlock_page(page);
 				ret = 0;
 			}
-			if (ret)
-				done = 1;
+			if (!err && ret < 0)
+				err = ret;
 
 			/*
 			 * the filesystem may choose to bump up nr_to_write.
@@ -3789,7 +3855,7 @@ retry:
 		pagevec_release(&pvec);
 		cond_resched();
 	}
-	if (!scanned && !done) {
+	if (!scanned && !done && !err) {
 		/*
 		 * We hit the last page and there is more work to be done: wrap
 		 * back to the start of the file
@@ -3799,7 +3865,7 @@ retry:
 		goto retry;
 	}
 	btrfs_add_delayed_iput(inode);
-	return ret;
+	return err;
 }
 
 static void flush_epd_write_bio(struct extent_page_data *epd)
@@ -4543,6 +4609,53 @@ struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info,
 	return NULL;
 }
 
+#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
+struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info,
+					       u64 start, unsigned long len)
+{
+	struct extent_buffer *eb, *exists = NULL;
+	int ret;
+
+	eb = find_extent_buffer(fs_info, start);
+	if (eb)
+		return eb;
+	eb = alloc_dummy_extent_buffer(start, len);
+	if (!eb)
+		return NULL;
+	eb->fs_info = fs_info;
+again:
+	ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
+	if (ret)
+		goto free_eb;
+	spin_lock(&fs_info->buffer_lock);
+	ret = radix_tree_insert(&fs_info->buffer_radix,
+				start >> PAGE_CACHE_SHIFT, eb);
+	spin_unlock(&fs_info->buffer_lock);
+	radix_tree_preload_end();
+	if (ret == -EEXIST) {
+		exists = find_extent_buffer(fs_info, start);
+		if (exists)
+			goto free_eb;
+		else
+			goto again;
+	}
+	check_buffer_tree_ref(eb);
+	set_bit(EXTENT_BUFFER_IN_TREE, &eb->bflags);
+
+	/*
+	 * We will free dummy extent buffer's if they come into
+	 * free_extent_buffer with a ref count of 2, but if we are using this we
+	 * want the buffers to stay in memory until we're done with them, so
+	 * bump the ref count again.
+	 */
+	atomic_inc(&eb->refs);
+	return eb;
+free_eb:
+	btrfs_release_extent_buffer(eb);
+	return exists;
+}
+#endif
+
 struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 					  u64 start, unsigned long len)
 {