Merge branch 'delalloc-buffer-write' into dev

Fix a bug in how we update i_disksize, and the error path in
inline_data_end.  Finally, drop an unnecessary creation of a journal
handle which was only needed for inline data, which can give us a
large performance gain in delayed allocation writes.

Signed-off-by: Theodore Ts'o <tytso@mit.edu>

3 files changed, 105 insertions, 179 deletions

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 98758e8ea7fc..2fa2dc6f6ac0 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -3603,9 +3603,6 @@ extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
 					   unsigned flags,
 					   struct page **pagep,
 					   void **fsdata);
-extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
-					 unsigned len, unsigned copied,
-					 struct page *page);
 extern int ext4_try_add_inline_entry(handle_t *handle,
 				     struct ext4_filename *fname,
 				     struct inode *dir, struct inode *inode);
diff --git a/fs/ext4/inline.c b/fs/ext4/inline.c
index 82bf4ff6be28..50a3031bf466 100644
--- a/fs/ext4/inline.c
+++ b/fs/ext4/inline.c
@@ -733,45 +733,83 @@ convert:
 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
 			       unsigned copied, struct page *page)
 {
-	int ret, no_expand;
+	handle_t *handle = ext4_journal_current_handle();
+	int no_expand;
 	void *kaddr;
 	struct ext4_iloc iloc;
+	int ret = 0, ret2;
+
+	if (unlikely(copied < len) && !PageUptodate(page))
+		copied = 0;
 
-	if (unlikely(copied < len)) {
-		if (!PageUptodate(page)) {
-			copied = 0;
+	if (likely(copied)) {
+		ret = ext4_get_inode_loc(inode, &iloc);
+		if (ret) {
+			unlock_page(page);
+			put_page(page);
+			ext4_std_error(inode->i_sb, ret);
 			goto out;
 		}
-	}
+		ext4_write_lock_xattr(inode, &no_expand);
+		BUG_ON(!ext4_has_inline_data(inode));
 
-	ret = ext4_get_inode_loc(inode, &iloc);
-	if (ret) {
-		ext4_std_error(inode->i_sb, ret);
-		copied = 0;
-		goto out;
-	}
+		/*
+		 * ei->i_inline_off may have changed since
+		 * ext4_write_begin() called
+		 * ext4_try_to_write_inline_data()
+		 */
+		(void) ext4_find_inline_data_nolock(inode);
 
-	ext4_write_lock_xattr(inode, &no_expand);
-	BUG_ON(!ext4_has_inline_data(inode));
+		kaddr = kmap_atomic(page);
+		ext4_write_inline_data(inode, &iloc, kaddr, pos, copied);
+		kunmap_atomic(kaddr);
+		SetPageUptodate(page);
+		/* clear page dirty so that writepages wouldn't work for us. */
+		ClearPageDirty(page);
 
-	/*
-	 * ei->i_inline_off may have changed since ext4_write_begin()
-	 * called ext4_try_to_write_inline_data()
-	 */
-	(void) ext4_find_inline_data_nolock(inode);
+		ext4_write_unlock_xattr(inode, &no_expand);
+		brelse(iloc.bh);
 
-	kaddr = kmap_atomic(page);
-	ext4_write_inline_data(inode, &iloc, kaddr, pos, len);
-	kunmap_atomic(kaddr);
-	SetPageUptodate(page);
-	/* clear page dirty so that writepages wouldn't work for us. */
-	ClearPageDirty(page);
+		/*
+		 * It's important to update i_size while still holding page
+		 * lock: page writeout could otherwise come in and zero
+		 * beyond i_size.
+		 */
+		ext4_update_inode_size(inode, pos + copied);
+	}
+	unlock_page(page);
+	put_page(page);
 
-	ext4_write_unlock_xattr(inode, &no_expand);
-	brelse(iloc.bh);
-	mark_inode_dirty(inode);
+	/*
+	 * Don't mark the inode dirty under page lock. First, it unnecessarily
+	 * makes the holding time of page lock longer. Second, it forces lock
+	 * ordering of page lock and transaction start for journaling
+	 * filesystems.
+	 */
+	if (likely(copied))
+		mark_inode_dirty(inode);
 out:
-	return copied;
+	/*
+	 * If we didn't copy as much data as expected, we need to trim back
+	 * size of xattr containing inline data.
+	 */
+	if (pos + len > inode->i_size && ext4_can_truncate(inode))
+		ext4_orphan_add(handle, inode);
+
+	ret2 = ext4_journal_stop(handle);
+	if (!ret)
+		ret = ret2;
+	if (pos + len > inode->i_size) {
+		ext4_truncate_failed_write(inode);
+		/*
+		 * If truncate failed early the inode might still be
+		 * on the orphan list; we need to make sure the inode
+		 * is removed from the orphan list in that case.
+		 */
+		if (inode->i_nlink)
+			ext4_orphan_del(NULL, inode);
+	}
+	return ret ? ret : copied;
 }
 
 struct buffer_head *
@@ -953,43 +991,6 @@ out:
 	return ret;
 }
 
-int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
-				  unsigned len, unsigned copied,
-				  struct page *page)
-{
-	int ret;
-
-	ret = ext4_write_inline_data_end(inode, pos, len, copied, page);
-	if (ret < 0) {
-		unlock_page(page);
-		put_page(page);
-		return ret;
-	}
-	copied = ret;
-
-	/*
-	 * No need to use i_size_read() here, the i_size
-	 * cannot change under us because we hold i_mutex.
-	 *
-	 * But it's important to update i_size while still holding page lock:
-	 * page writeout could otherwise come in and zero beyond i_size.
-	 */
-	if (pos+copied > inode->i_size)
-		i_size_write(inode, pos+copied);
-	unlock_page(page);
-	put_page(page);
-
-	/*
-	 * Don't mark the inode dirty under page lock. First, it unnecessarily
-	 * makes the holding time of page lock longer. Second, it forces lock
-	 * ordering of page lock and transaction start for journaling
-	 * filesystems.
-	 */
-	mark_inode_dirty(inode);
-
-	return copied;
-}
-
 #ifdef INLINE_DIR_DEBUG
 void ext4_show_inline_dir(struct inode *dir, struct buffer_head *bh,
 			  void *inline_start, int inline_size)
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 62e9165bc69c..8204176256c8 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1284,22 +1284,14 @@ static int ext4_write_end(struct file *file,
 	loff_t old_size = inode->i_size;
 	int ret = 0, ret2;
 	int i_size_changed = 0;
-	int inline_data = ext4_has_inline_data(inode);
 	bool verity = ext4_verity_in_progress(inode);
 
 	trace_ext4_write_end(inode, pos, len, copied);
-	if (inline_data) {
-		ret = ext4_write_inline_data_end(inode, pos, len,
-						 copied, page);
-		if (ret < 0) {
-			unlock_page(page);
-			put_page(page);
-			goto errout;
-		}
-		copied = ret;
-	} else
-		copied = block_write_end(file, mapping, pos,
-					 len, copied, page, fsdata);
+
+	if (ext4_has_inline_data(inode))
+		return ext4_write_inline_data_end(inode, pos, len, copied, page);
+
+	copied = block_write_end(file, mapping, pos, len, copied, page, fsdata);
 	/*
 	 * it's important to update i_size while still holding page lock:
 	 * page writeout could otherwise come in and zero beyond i_size.
@@ -1320,7 +1312,7 @@ static int ext4_write_end(struct file *file,
 	 * ordering of page lock and transaction start for journaling
 	 * filesystems.
 	 */
-	if (i_size_changed || inline_data)
+	if (i_size_changed)
 		ret = ext4_mark_inode_dirty(handle, inode);
 
 	if (pos + len > inode->i_size && !verity && ext4_can_truncate(inode))
@@ -1329,7 +1321,7 @@ static int ext4_write_end(struct file *file,
 		 * inode->i_size. So truncate them
 		 */
 		ext4_orphan_add(handle, inode);
-errout:
+
 	ret2 = ext4_journal_stop(handle);
 	if (!ret)
 		ret = ret2;
@@ -1395,7 +1387,6 @@ static int ext4_journalled_write_end(struct file *file,
 	int partial = 0;
 	unsigned from, to;
 	int size_changed = 0;
-	int inline_data = ext4_has_inline_data(inode);
 	bool verity = ext4_verity_in_progress(inode);
 
 	trace_ext4_journalled_write_end(inode, pos, len, copied);
@@ -1404,16 +1395,10 @@ static int ext4_journalled_write_end(struct file *file,
 
 	BUG_ON(!ext4_handle_valid(handle));
 
-	if (inline_data) {
-		ret = ext4_write_inline_data_end(inode, pos, len,
-						 copied, page);
-		if (ret < 0) {
-			unlock_page(page);
-			put_page(page);
-			goto errout;
-		}
-		copied = ret;
-	} else if (unlikely(copied < len) && !PageUptodate(page)) {
+	if (ext4_has_inline_data(inode))
+		return ext4_write_inline_data_end(inode, pos, len, copied, page);
+
+	if (unlikely(copied < len) && !PageUptodate(page)) {
 		copied = 0;
 		ext4_journalled_zero_new_buffers(handle, inode, page, from, to);
 	} else {
@@ -1436,7 +1421,7 @@ static int ext4_journalled_write_end(struct file *file,
 	if (old_size < pos && !verity)
 		pagecache_isize_extended(inode, old_size, pos);
 
-	if (size_changed || inline_data) {
+	if (size_changed) {
 		ret2 = ext4_mark_inode_dirty(handle, inode);
 		if (!ret)
 			ret = ret2;
@@ -1449,7 +1434,6 @@ static int ext4_journalled_write_end(struct file *file,
 		 */
 		ext4_orphan_add(handle, inode);
 
-errout:
 	ret2 = ext4_journal_stop(handle);
 	if (!ret)
 		ret = ret2;
@@ -2932,19 +2916,6 @@ static int ext4_nonda_switch(struct super_block *sb)
 	return 0;
 }
 
-/* We always reserve for an inode update; the superblock could be there too */
-static int ext4_da_write_credits(struct inode *inode, loff_t pos, unsigned len)
-{
-	if (likely(ext4_has_feature_large_file(inode->i_sb)))
-		return 1;
-
-	if (pos + len <= 0x7fffffffULL)
-		return 1;
-
-	/* We might need to update the superblock to set LARGE_FILE */
-	return 2;
-}
-
 static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
 			       loff_t pos, unsigned len, unsigned flags,
 			       struct page **pagep, void **fsdata)
@@ -2953,7 +2924,6 @@ static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
 	struct page *page;
 	pgoff_t index;
 	struct inode *inode = mapping->host;
-	handle_t *handle;
 
 	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
 		return -EIO;
@@ -2979,41 +2949,11 @@ static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
 			return 0;
 	}
 
-	/*
-	 * grab_cache_page_write_begin() can take a long time if the
-	 * system is thrashing due to memory pressure, or if the page
-	 * is being written back.  So grab it first before we start
-	 * the transaction handle.  This also allows us to allocate
-	 * the page (if needed) without using GFP_NOFS.
-	 */
-retry_grab:
+retry:
 	page = grab_cache_page_write_begin(mapping, index, flags);
 	if (!page)
 		return -ENOMEM;
-	unlock_page(page);
 
-	/*
-	 * With delayed allocation, we don't log the i_disksize update
-	 * if there is delayed block allocation. But we still need
-	 * to journalling the i_disksize update if writes to the end
-	 * of file which has an already mapped buffer.
-	 */
-retry_journal:
-	handle = ext4_journal_start(inode, EXT4_HT_WRITE_PAGE,
-				ext4_da_write_credits(inode, pos, len));
-	if (IS_ERR(handle)) {
-		put_page(page);
-		return PTR_ERR(handle);
-	}
-
-	lock_page(page);
-	if (page->mapping != mapping) {
-		/* The page got truncated from under us */
-		unlock_page(page);
-		put_page(page);
-		ext4_journal_stop(handle);
-		goto retry_grab;
-	}
 	/* In case writeback began while the page was unlocked */
 	wait_for_stable_page(page);
 
@@ -3025,20 +2965,18 @@ retry_journal:
 #endif
 	if (ret < 0) {
 		unlock_page(page);
-		ext4_journal_stop(handle);
+		put_page(page);
 		/*
 		 * block_write_begin may have instantiated a few blocks
 		 * outside i_size.  Trim these off again. Don't need
-		 * i_size_read because we hold i_mutex.
+		 * i_size_read because we hold inode lock.
 		 */
 		if (pos + len > inode->i_size)
 			ext4_truncate_failed_write(inode);
 
 		if (ret == -ENOSPC &&
 		    ext4_should_retry_alloc(inode->i_sb, &retries))
-			goto retry_journal;
-
-		put_page(page);
+			goto retry;
 		return ret;
 	}
 
@@ -3075,8 +3013,6 @@ static int ext4_da_write_end(struct file *file,
 			     struct page *page, void *fsdata)
 {
 	struct inode *inode = mapping->host;
-	int ret = 0, ret2;
-	handle_t *handle = ext4_journal_current_handle();
 	loff_t new_i_size;
 	unsigned long start, end;
 	int write_mode = (int)(unsigned long)fsdata;
@@ -3086,44 +3022,36 @@ static int ext4_da_write_end(struct file *file,
 				      len, copied, page, fsdata);
 
 	trace_ext4_da_write_end(inode, pos, len, copied);
+
+	if (write_mode != CONVERT_INLINE_DATA &&
+	    ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA) &&
+	    ext4_has_inline_data(inode))
+		return ext4_write_inline_data_end(inode, pos, len, copied, page);
+
 	start = pos & (PAGE_SIZE - 1);
 	end = start + copied - 1;
 
 	/*
-	 * generic_write_end() will run mark_inode_dirty() if i_size
-	 * changes.  So let's piggyback the i_disksize mark_inode_dirty
-	 * into that.
+	 * Since we are holding inode lock, we are sure i_disksize <=
+	 * i_size. We also know that if i_disksize < i_size, there are
+	 * delalloc writes pending in the range upto i_size. If the end of
+	 * the current write is <= i_size, there's no need to touch
+	 * i_disksize since writeback will push i_disksize upto i_size
+	 * eventually. If the end of the current write is > i_size and
+	 * inside an allocated block (ext4_da_should_update_i_disksize()
+	 * check), we need to update i_disksize here as neither
+	 * ext4_writepage() nor certain ext4_writepages() paths not
+	 * allocating blocks update i_disksize.
+	 *
+	 * Note that we defer inode dirtying to generic_write_end() /
+	 * ext4_da_write_inline_data_end().
 	 */
 	new_i_size = pos + copied;
-	if (copied && new_i_size > EXT4_I(inode)->i_disksize) {
-		if (ext4_has_inline_data(inode) ||
-		    ext4_da_should_update_i_disksize(page, end)) {
-			ext4_update_i_disksize(inode, new_i_size);
-			/* We need to mark inode dirty even if
-			 * new_i_size is less that inode->i_size
-			 * bu greater than i_disksize.(hint delalloc)
-			 */
-			ret = ext4_mark_inode_dirty(handle, inode);
-		}
-	}
+	if (copied && new_i_size > inode->i_size &&
+	    ext4_da_should_update_i_disksize(page, end))
+		ext4_update_i_disksize(inode, new_i_size);
 
-	if (write_mode != CONVERT_INLINE_DATA &&
-	    ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA) &&
-	    ext4_has_inline_data(inode))
-		ret2 = ext4_da_write_inline_data_end(inode, pos, len, copied,
-						     page);
-	else
-		ret2 = generic_write_end(file, mapping, pos, len, copied,
-							page, fsdata);
-
-	copied = ret2;
-	if (ret2 < 0)
-		ret = ret2;
-	ret2 = ext4_journal_stop(handle);
-	if (unlikely(ret2 && !ret))
-		ret = ret2;
-
-	return ret ? ret : copied;
+	return generic_write_end(file, mapping, pos, len, copied, page, fsdata);
 }
 
 /*