ext4: refactor punch hole code

Move common code in ext4_ind_punch_hole() and ext4_ext_punch_hole()
into ext4_punch_hole().  This saves over 150 lines of code.

This also fixes a potential bug when the punch_hole() code is racing
against indirect-to-extents or extents-to-indirect migation.  We are
currently using i_mutex to protect against changes to the inode flag;
specifically, the append-only, immutable, and extents inode flags.  So
we need to take i_mutex before deciding whether to use the
extents-specific or indirect-specific punch_hole code.

Also, there was a missing call to ext4_inode_block_unlocked_dio() in
the indirect punch codepath.  This was added in commit 02d262df
to block DIO readers racing against the punch operation in the
codepath for extent-mapped inodes, but it was missing for
indirect-block mapped inodes.  One of the advantages of refactoring
the code is that it makes such oversights much less likely.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

fs/ext4/ext4.h

@@ -2110,7 +2110,8 @@ extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
 extern void ext4_ind_truncate(struct inode *inode);
-extern int ext4_ind_punch_hole(struct file *file, loff_t offset, loff_t length);
+extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
+				 ext4_lblk_t first, ext4_lblk_t stop);
 
 /* ioctl.c */
 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
@@ -2575,8 +2576,8 @@ extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 			       struct ext4_map_blocks *map, int flags);
 extern void ext4_ext_truncate(struct inode *);
-extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
-				loff_t length);
+extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
+				 ext4_lblk_t end);
 extern void ext4_ext_init(struct super_block *);
 extern void ext4_ext_release(struct super_block *);
 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,

fs/ext4/extents.c

@@ -2599,8 +2599,8 @@ ext4_ext_more_to_rm(struct ext4_ext_path *path)
 	return 1;
 }
 
-static int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
-				 ext4_lblk_t end)
+int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
+			  ext4_lblk_t end)
 {
 	struct super_block *sb = inode->i_sb;
 	int depth = ext_depth(inode);
@@ -4623,187 +4623,6 @@ static int ext4_xattr_fiemap(struct inode *inode,
 	return (error < 0 ? error : 0);
 }
 
-/*
- * ext4_ext_punch_hole
- *
- * Punches a hole of "length" bytes in a file starting
- * at byte "offset"
- *
- * @inode:  The inode of the file to punch a hole in
- * @offset: The starting byte offset of the hole
- * @length: The length of the hole
- *
- * Returns the number of blocks removed or negative on err
- */
-int ext4_ext_punch_hole(struct file *file, loff_t offset, loff_t length)
-{
-	struct inode *inode = file_inode(file);
-	struct super_block *sb = inode->i_sb;
-	ext4_lblk_t first_block, stop_block;
-	struct address_space *mapping = inode->i_mapping;
-	handle_t *handle;
-	loff_t first_page, last_page, page_len;
-	loff_t first_page_offset, last_page_offset;
-	int credits, err = 0;
-
-	/*
-	 * Write out all dirty pages to avoid race conditions
-	 * Then release them.
-	 */
-	if (mapping->nrpages && mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
-		err = filemap_write_and_wait_range(mapping,
-			offset, offset + length - 1);
-
-		if (err)
-			return err;
-	}
-
-	mutex_lock(&inode->i_mutex);
-	/* It's not possible punch hole on append only file */
-	if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) {
-		err = -EPERM;
-		goto out_mutex;
-	}
-	if (IS_SWAPFILE(inode)) {
-		err = -ETXTBSY;
-		goto out_mutex;
-	}
-
-	/* No need to punch hole beyond i_size */
-	if (offset >= inode->i_size)
-		goto out_mutex;
-
-	/*
-	 * If the hole extends beyond i_size, set the hole
-	 * to end after the page that contains i_size
-	 */
-	if (offset + length > inode->i_size) {
-		length = inode->i_size +
-		   PAGE_CACHE_SIZE - (inode->i_size & (PAGE_CACHE_SIZE - 1)) -
-		   offset;
-	}
-
-	first_page = (offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
-	last_page = (offset + length) >> PAGE_CACHE_SHIFT;
-
-	first_page_offset = first_page << PAGE_CACHE_SHIFT;
-	last_page_offset = last_page << PAGE_CACHE_SHIFT;
-
-	/* Now release the pages */
-	if (last_page_offset > first_page_offset) {
-		truncate_pagecache_range(inode, first_page_offset,
-					 last_page_offset - 1);
-	}
-
-	/* Wait all existing dio workers, newcomers will block on i_mutex */
-	ext4_inode_block_unlocked_dio(inode);
-	err = ext4_flush_unwritten_io(inode);
-	if (err)
-		goto out_dio;
-	inode_dio_wait(inode);
-
-	credits = ext4_writepage_trans_blocks(inode);
-	handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, credits);
-	if (IS_ERR(handle)) {
-		err = PTR_ERR(handle);
-		goto out_dio;
-	}
-
-
-	/*
-	 * Now we need to zero out the non-page-aligned data in the
-	 * pages at the start and tail of the hole, and unmap the buffer
-	 * heads for the block aligned regions of the page that were
-	 * completely zeroed.
-	 */
-	if (first_page > last_page) {
-		/*
-		 * If the file space being truncated is contained within a page
-		 * just zero out and unmap the middle of that page
-		 */
-		err = ext4_discard_partial_page_buffers(handle,
-			mapping, offset, length, 0);
-
-		if (err)
-			goto out;
-	} else {
-		/*
-		 * zero out and unmap the partial page that contains
-		 * the start of the hole
-		 */
-		page_len  = first_page_offset - offset;
-		if (page_len > 0) {
-			err = ext4_discard_partial_page_buffers(handle, mapping,
-						   offset, page_len, 0);
-			if (err)
-				goto out;
-		}
-
-		/*
-		 * zero out and unmap the partial page that contains
-		 * the end of the hole
-		 */
-		page_len = offset + length - last_page_offset;
-		if (page_len > 0) {
-			err = ext4_discard_partial_page_buffers(handle, mapping,
-					last_page_offset, page_len, 0);
-			if (err)
-				goto out;
-		}
-	}
-
-	/*
-	 * If i_size is contained in the last page, we need to
-	 * unmap and zero the partial page after i_size
-	 */
-	if (inode->i_size >> PAGE_CACHE_SHIFT == last_page &&
-	   inode->i_size % PAGE_CACHE_SIZE != 0) {
-
-		page_len = PAGE_CACHE_SIZE -
-			(inode->i_size & (PAGE_CACHE_SIZE - 1));
-
-		if (page_len > 0) {
-			err = ext4_discard_partial_page_buffers(handle,
-			  mapping, inode->i_size, page_len, 0);
-
-			if (err)
-				goto out;
-		}
-	}
-
-	first_block = (offset + sb->s_blocksize - 1) >>
-		EXT4_BLOCK_SIZE_BITS(sb);
-	stop_block = (offset + length) >> EXT4_BLOCK_SIZE_BITS(sb);
-
-	/* If there are no blocks to remove, return now */
-	if (first_block >= stop_block)
-		goto out;
-
-	down_write(&EXT4_I(inode)->i_data_sem);
-	ext4_discard_preallocations(inode);
-
-	err = ext4_es_remove_extent(inode, first_block,
-				    stop_block - first_block);
-	err = ext4_ext_remove_space(inode, first_block, stop_block - 1);
-
-	ext4_discard_preallocations(inode);
-
-	if (IS_SYNC(inode))
-		ext4_handle_sync(handle);
-
-	up_write(&EXT4_I(inode)->i_data_sem);
-
-out:
-	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
-	ext4_mark_inode_dirty(handle, inode);
-	ext4_journal_stop(handle);
-out_dio:
-	ext4_inode_resume_unlocked_dio(inode);
-out_mutex:
-	mutex_unlock(&inode->i_mutex);
-	return err;
-}
-
 int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 		__u64 start, __u64 len)
 {

fs/ext4/indirect.c

@@ -1434,8 +1434,8 @@ static int free_hole_blocks(handle_t *handle, struct inode *inode,
 	return ret;
 }
 
-static int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
-				 ext4_lblk_t first, ext4_lblk_t stop)
+int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
+			  ext4_lblk_t first, ext4_lblk_t stop)
 {
 	int addr_per_block = EXT4_ADDR_PER_BLOCK(inode->i_sb);
 	int level, ret = 0;
@@ -1469,157 +1469,3 @@ static int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
 	return ret;
 }
 
-int ext4_ind_punch_hole(struct file *file, loff_t offset, loff_t length)
-{
-	struct inode *inode = file_inode(file);
-	struct super_block *sb = inode->i_sb;
-	ext4_lblk_t first_block, stop_block;
-	struct address_space *mapping = inode->i_mapping;
-	handle_t *handle = NULL;
-	loff_t first_page, last_page, page_len;
-	loff_t first_page_offset, last_page_offset;
-	int err = 0;
-
-	/*
-	 * Write out all dirty pages to avoid race conditions
-	 * Then release them.
-	 */
-	if (mapping->nrpages && mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
-		err = filemap_write_and_wait_range(mapping,
-			offset, offset + length - 1);
-		if (err)
-			return err;
-	}
-
-	mutex_lock(&inode->i_mutex);
-	/* It's not possible punch hole on append only file */
-	if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) {
-		err = -EPERM;
-		goto out_mutex;
-	}
-	if (IS_SWAPFILE(inode)) {
-		err = -ETXTBSY;
-		goto out_mutex;
-	}
-
-	/* No need to punch hole beyond i_size */
-	if (offset >= inode->i_size)
-		goto out_mutex;
-
-	/*
-	 * If the hole extents beyond i_size, set the hole
-	 * to end after the page that contains i_size
-	 */
-	if (offset + length > inode->i_size) {
-		length = inode->i_size +
-		    PAGE_CACHE_SIZE - (inode->i_size & (PAGE_CACHE_SIZE - 1)) -
-		    offset;
-	}
-
-	first_page = (offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
-	last_page = (offset + length) >> PAGE_CACHE_SHIFT;
-
-	first_page_offset = first_page << PAGE_CACHE_SHIFT;
-	last_page_offset = last_page << PAGE_CACHE_SHIFT;
-
-	/* Now release the pages */
-	if (last_page_offset > first_page_offset) {
-		truncate_pagecache_range(inode, first_page_offset,
-					 last_page_offset - 1);
-	}
-
-	/* Wait all existing dio works, newcomers will block on i_mutex */
-	inode_dio_wait(inode);
-
-	handle = start_transaction(inode);
-	if (IS_ERR(handle))
-		goto out_mutex;
-
-	/*
-	 * Now we need to zero out the non-page-aligned data in the
-	 * pages at the start and tail of the hole, and unmap the buffer
-	 * heads for the block aligned regions of the page that were
-	 * completely zerod.
-	 */
-	if (first_page > last_page) {
-		/*
-		 * If the file space being truncated is contained within a page
-		 * just zero out and unmap the middle of that page
-		 */
-		err = ext4_discard_partial_page_buffers(handle,
-			mapping, offset, length, 0);
-		if (err)
-			goto out;
-	} else {
-		/*
-		 * Zero out and unmap the paritial page that contains
-		 * the start of the hole
-		 */
-		page_len = first_page_offset - offset;
-		if (page_len > 0) {
-			err = ext4_discard_partial_page_buffers(handle, mapping,
-							offset, page_len, 0);
-			if (err)
-				goto out;
-		}
-
-		/*
-		 * Zero out and unmap the partial page that contains
-		 * the end of the hole
-		 */
-		page_len = offset + length - last_page_offset;
-		if (page_len > 0) {
-			err = ext4_discard_partial_page_buffers(handle, mapping,
-						last_page_offset, page_len, 0);
-			if (err)
-				goto out;
-		}
-	}
-
-	/*
-	 * If i_size contained in the last page, we need to
-	 * unmap and zero the paritial page after i_size
-	 */
-	if (inode->i_size >> PAGE_CACHE_SHIFT == last_page &&
-	    inode->i_size % PAGE_CACHE_SIZE != 0) {
-		page_len = PAGE_CACHE_SIZE -
-			(inode->i_size & (PAGE_CACHE_SIZE - 1));
-		if (page_len > 0) {
-			err = ext4_discard_partial_page_buffers(handle,
-				mapping, inode->i_size, page_len, 0);
-			if (err)
-				goto out;
-		}
-	}
-
-	first_block = (offset + sb->s_blocksize - 1) >>
-		EXT4_BLOCK_SIZE_BITS(sb);
-	stop_block = (offset + length) >> EXT4_BLOCK_SIZE_BITS(sb);
-
-	if (first_block >= stop_block)
-		goto out;
-
-	down_write(&EXT4_I(inode)->i_data_sem);
-	ext4_discard_preallocations(inode);
-
-	err = ext4_es_remove_extent(inode, first_block,
-				    stop_block - first_block);
-	err = ext4_free_hole_blocks(handle, inode, first_block, stop_block);
-
-	ext4_discard_preallocations(inode);
-
-	if (IS_SYNC(inode))
-		ext4_handle_sync(handle);
-
-	up_write(&EXT4_I(inode)->i_data_sem);
-
-out:
-	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
-	ext4_mark_inode_dirty(handle, inode);
-	ext4_journal_stop(handle);
-
-out_mutex:
-	mutex_unlock(&inode->i_mutex);
-
-	return err;
-}

fs/ext4/inode.c

@@ -3566,20 +3566,190 @@ int ext4_can_truncate(struct inode *inode)
 int ext4_punch_hole(struct file *file, loff_t offset, loff_t length)
 {
 	struct inode *inode = file_inode(file);
+	struct super_block *sb = inode->i_sb;
+	ext4_lblk_t first_block, stop_block;
+	struct address_space *mapping = inode->i_mapping;
+	loff_t first_page, last_page, page_len;
+	loff_t first_page_offset, last_page_offset;
+	handle_t *handle;
+	unsigned int credits;
+	int ret = 0;
+
 	if (!S_ISREG(inode->i_mode))
 		return -EOPNOTSUPP;
 
-	if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
-		return ext4_ind_punch_hole(file, offset, length);
-
-	if (EXT4_SB(inode->i_sb)->s_cluster_ratio > 1) {
+	if (EXT4_SB(sb)->s_cluster_ratio > 1) {
 		/* TODO: Add support for bigalloc file systems */
 		return -EOPNOTSUPP;
 	}
 
 	trace_ext4_punch_hole(inode, offset, length);
 
-	return ext4_ext_punch_hole(file, offset, length);
+	/*
+	 * Write out all dirty pages to avoid race conditions
+	 * Then release them.
+	 */
+	if (mapping->nrpages && mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
+		ret = filemap_write_and_wait_range(mapping, offset,
+						   offset + length - 1);
+		if (ret)
+			return ret;
+	}
+
+	mutex_lock(&inode->i_mutex);
+	/* It's not possible punch hole on append only file */
+	if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) {
+		ret = -EPERM;
+		goto out_mutex;
+	}
+	if (IS_SWAPFILE(inode)) {
+		ret = -ETXTBSY;
+		goto out_mutex;
+	}
+
+	/* No need to punch hole beyond i_size */
+	if (offset >= inode->i_size)
+		goto out_mutex;
+
+	/*
+	 * If the hole extends beyond i_size, set the hole
+	 * to end after the page that contains i_size
+	 */
+	if (offset + length > inode->i_size) {
+		length = inode->i_size +
+		   PAGE_CACHE_SIZE - (inode->i_size & (PAGE_CACHE_SIZE - 1)) -
+		   offset;
+	}
+
+	first_page = (offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
+	last_page = (offset + length) >> PAGE_CACHE_SHIFT;
+
+	first_page_offset = first_page << PAGE_CACHE_SHIFT;
+	last_page_offset = last_page << PAGE_CACHE_SHIFT;
+
+	/* Now release the pages */
+	if (last_page_offset > first_page_offset) {
+		truncate_pagecache_range(inode, first_page_offset,
+					 last_page_offset - 1);
+	}
+
+	/* Wait all existing dio workers, newcomers will block on i_mutex */
+	ext4_inode_block_unlocked_dio(inode);
+	ret = ext4_flush_unwritten_io(inode);
+	if (ret)
+		goto out_dio;
+	inode_dio_wait(inode);
+
+	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
+		credits = ext4_writepage_trans_blocks(inode);
+	else
+		credits = ext4_blocks_for_truncate(inode);
+	handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, credits);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		ext4_std_error(sb, ret);
+		goto out_dio;
+	}
+
+	/*
+	 * Now we need to zero out the non-page-aligned data in the
+	 * pages at the start and tail of the hole, and unmap the
+	 * buffer heads for the block aligned regions of the page that
+	 * were completely zeroed.
+	 */
+	if (first_page > last_page) {
+		/*
+		 * If the file space being truncated is contained
+		 * within a page just zero out and unmap the middle of
+		 * that page
+		 */
+		ret = ext4_discard_partial_page_buffers(handle,
+			mapping, offset, length, 0);
+
+		if (ret)
+			goto out_stop;
+	} else {
+		/*
+		 * zero out and unmap the partial page that contains
+		 * the start of the hole
+		 */
+		page_len = first_page_offset - offset;
+		if (page_len > 0) {
+			ret = ext4_discard_partial_page_buffers(handle, mapping,
+						offset, page_len, 0);
+			if (ret)
+				goto out_stop;
+		}
+
+		/*
+		 * zero out and unmap the partial page that contains
+		 * the end of the hole
+		 */
+		page_len = offset + length - last_page_offset;
+		if (page_len > 0) {
+			ret = ext4_discard_partial_page_buffers(handle, mapping,
+					last_page_offset, page_len, 0);
+			if (ret)
+				goto out_stop;
+		}
+	}
+
+	/*
+	 * If i_size is contained in the last page, we need to
+	 * unmap and zero the partial page after i_size
+	 */
+	if (inode->i_size >> PAGE_CACHE_SHIFT == last_page &&
+	   inode->i_size % PAGE_CACHE_SIZE != 0) {
+		page_len = PAGE_CACHE_SIZE -
+			(inode->i_size & (PAGE_CACHE_SIZE - 1));
+
+		if (page_len > 0) {
+			ret = ext4_discard_partial_page_buffers(handle,
+					mapping, inode->i_size, page_len, 0);
+
+			if (ret)
+				goto out_stop;
+		}
+	}
+
+	first_block = (offset + sb->s_blocksize - 1) >>
+		EXT4_BLOCK_SIZE_BITS(sb);
+	stop_block = (offset + length) >> EXT4_BLOCK_SIZE_BITS(sb);
+
+	/* If there are no blocks to remove, return now */
+	if (first_block >= stop_block)
+		goto out_stop;
+
+	down_write(&EXT4_I(inode)->i_data_sem);
+	ext4_discard_preallocations(inode);
+
+	ret = ext4_es_remove_extent(inode, first_block,
+				    stop_block - first_block);
+	if (ret) {
+		up_write(&EXT4_I(inode)->i_data_sem);
+		goto out_stop;
+	}
+
+	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
+		ret = ext4_ext_remove_space(inode, first_block,
+					    stop_block - 1);
+	else
+		ret = ext4_free_hole_blocks(handle, inode, first_block,
+					    stop_block);
+
+	ext4_discard_preallocations(inode);
+	if (IS_SYNC(inode))
+		ext4_handle_sync(handle);
+	up_write(&EXT4_I(inode)->i_data_sem);
+	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
+	ext4_mark_inode_dirty(handle, inode);
+out_stop:
+	ext4_journal_stop(handle);
+out_dio:
+	ext4_inode_resume_unlocked_dio(inode);
+out_mutex:
+	mutex_unlock(&inode->i_mutex);
+	return ret;
 }
 
 /*