Merge tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4

Pull ext4 bugfixes from Ted Ts'o:
 "Ext4 bug fixes for 3.17, to provide better handling of memory
  allocation failures, and to fix some journaling bugs involving
  journal checksums and FALLOC_FL_ZERO_RANGE"

* tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: fix same-dir rename when inline data directory overflows
  jbd2: fix descriptor block size handling errors with journal_csum
  jbd2: fix infinite loop when recovering corrupt journal blocks
  ext4: update i_disksize coherently with block allocation on error path
  ext4: fix transaction issues for ext4_fallocate and ext_zero_range
  ext4: fix incorect journal credits reservation in ext4_zero_range
  ext4: move i_size,i_disksize update routines to helper function
  ext4: fix BUG_ON in mb_free_blocks()
  ext4: propagate errors up to ext4_find_entry()'s callers

fs/ext4/ext4.h

@@ -1825,7 +1825,7 @@ ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
 /*
  * Special error return code only used by dx_probe() and its callers.
  */
-#define ERR_BAD_DX_DIR	-75000
+#define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
 
 /*
  * Timeout and state flag for lazy initialization inode thread.
@@ -2454,6 +2454,22 @@ static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
 	up_write(&EXT4_I(inode)->i_data_sem);
 }
 
+/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
+static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
+{
+	int changed = 0;
+
+	if (newsize > inode->i_size) {
+		i_size_write(inode, newsize);
+		changed = 1;
+	}
+	if (newsize > EXT4_I(inode)->i_disksize) {
+		ext4_update_i_disksize(inode, newsize);
+		changed |= 2;
+	}
+	return changed;
+}
+
 struct ext4_group_info {
 	unsigned long   bb_state;
 	struct rb_root  bb_free_root;

fs/ext4/extents.c

@@ -4665,7 +4665,8 @@ void ext4_ext_truncate(handle_t *handle, struct inode *inode)
 }
 
 static int ext4_alloc_file_blocks(struct file *file, ext4_lblk_t offset,
-				  ext4_lblk_t len, int flags, int mode)
+				  ext4_lblk_t len, loff_t new_size,
+				  int flags, int mode)
 {
 	struct inode *inode = file_inode(file);
 	handle_t *handle;
@@ -4674,8 +4675,10 @@ static int ext4_alloc_file_blocks(struct file *file, ext4_lblk_t offset,
 	int retries = 0;
 	struct ext4_map_blocks map;
 	unsigned int credits;
+	loff_t epos;
 
 	map.m_lblk = offset;
+	map.m_len = len;
 	/*
 	 * Don't normalize the request if it can fit in one extent so
 	 * that it doesn't get unnecessarily split into multiple
@@ -4690,9 +4693,7 @@ static int ext4_alloc_file_blocks(struct file *file, ext4_lblk_t offset,
 	credits = ext4_chunk_trans_blocks(inode, len);
 
 retry:
-	while (ret >= 0 && ret < len) {
-		map.m_lblk = map.m_lblk + ret;
-		map.m_len = len = len - ret;
+	while (ret >= 0 && len) {
 		handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS,
 					    credits);
 		if (IS_ERR(handle)) {
@@ -4709,6 +4710,21 @@ static int ext4_alloc_file_blocks(struct file *file, ext4_lblk_t offset,
 			ret2 = ext4_journal_stop(handle);
 			break;
 		}
+		map.m_lblk += ret;
+		map.m_len = len = len - ret;
+		epos = (loff_t)map.m_lblk << inode->i_blkbits;
+		inode->i_ctime = ext4_current_time(inode);
+		if (new_size) {
+			if (epos > new_size)
+				epos = new_size;
+			if (ext4_update_inode_size(inode, epos) & 0x1)
+				inode->i_mtime = inode->i_ctime;
+		} else {
+			if (epos > inode->i_size)
+				ext4_set_inode_flag(inode,
+						    EXT4_INODE_EOFBLOCKS);
+		}
+		ext4_mark_inode_dirty(handle, inode);
 		ret2 = ext4_journal_stop(handle);
 		if (ret2)
 			break;
@@ -4731,7 +4747,8 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 	loff_t new_size = 0;
 	int ret = 0;
 	int flags;
-	int partial;
+	int credits;
+	int partial_begin, partial_end;
 	loff_t start, end;
 	ext4_lblk_t lblk;
 	struct address_space *mapping = inode->i_mapping;
@@ -4771,7 +4788,8 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 
 	if (start < offset || end > offset + len)
 		return -EINVAL;
-	partial = (offset + len) & ((1 << blkbits) - 1);
+	partial_begin = offset & ((1 << blkbits) - 1);
+	partial_end = (offset + len) & ((1 << blkbits) - 1);
 
 	lblk = start >> blkbits;
 	max_blocks = (end >> blkbits);
@@ -4805,7 +4823,7 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 		 * If we have a partial block after EOF we have to allocate
 		 * the entire block.
 		 */
-		if (partial)
+		if (partial_end)
 			max_blocks += 1;
 	}
 
@@ -4813,6 +4831,7 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 
 		/* Now release the pages and zero block aligned part of pages*/
 		truncate_pagecache_range(inode, start, end - 1);
+		inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
 
 		/* Wait all existing dio workers, newcomers will block on i_mutex */
 		ext4_inode_block_unlocked_dio(inode);
@@ -4825,13 +4844,22 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 		if (ret)
 			goto out_dio;
 
-		ret = ext4_alloc_file_blocks(file, lblk, max_blocks, flags,
-					     mode);
+		ret = ext4_alloc_file_blocks(file, lblk, max_blocks, new_size,
+					     flags, mode);
 		if (ret)
 			goto out_dio;
 	}
+	if (!partial_begin && !partial_end)
+		goto out_dio;
 
-	handle = ext4_journal_start(inode, EXT4_HT_MISC, 4);
+	/*
+	 * In worst case we have to writeout two nonadjacent unwritten
+	 * blocks and update the inode
+	 */
+	credits = (2 * ext4_ext_index_trans_blocks(inode, 2)) + 1;
+	if (ext4_should_journal_data(inode))
+		credits += 2;
+	handle = ext4_journal_start(inode, EXT4_HT_MISC, credits);
 	if (IS_ERR(handle)) {
 		ret = PTR_ERR(handle);
 		ext4_std_error(inode->i_sb, ret);
@@ -4839,12 +4867,8 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 	}
 
 	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
-
 	if (new_size) {
-		if (new_size > i_size_read(inode))
-			i_size_write(inode, new_size);
-		if (new_size > EXT4_I(inode)->i_disksize)
-			ext4_update_i_disksize(inode, new_size);
+		ext4_update_inode_size(inode, new_size);
 	} else {
 		/*
 		* Mark that we allocate beyond EOF so the subsequent truncate
@@ -4853,7 +4877,6 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 		if ((offset + len) > i_size_read(inode))
 			ext4_set_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
 	}
-
 	ext4_mark_inode_dirty(handle, inode);
 
 	/* Zero out partial block at the edges of the range */
@@ -4880,13 +4903,11 @@ static long ext4_zero_range(struct file *file, loff_t offset,
 long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
 {
 	struct inode *inode = file_inode(file);
-	handle_t *handle;
 	loff_t new_size = 0;
 	unsigned int max_blocks;
 	int ret = 0;
 	int flags;
 	ext4_lblk_t lblk;
-	struct timespec tv;
 	unsigned int blkbits = inode->i_blkbits;
 
 	/* Return error if mode is not supported */
@@ -4937,36 +4958,15 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
 			goto out;
 	}
 
-	ret = ext4_alloc_file_blocks(file, lblk, max_blocks, flags, mode);
+	ret = ext4_alloc_file_blocks(file, lblk, max_blocks, new_size,
+				     flags, mode);
 	if (ret)
 		goto out;
 
-	handle = ext4_journal_start(inode, EXT4_HT_INODE, 2);
-	if (IS_ERR(handle))
-		goto out;
-
-	tv = inode->i_ctime = ext4_current_time(inode);
-
-	if (new_size) {
-		if (new_size > i_size_read(inode)) {
-			i_size_write(inode, new_size);
-			inode->i_mtime = tv;
-		}
-		if (new_size > EXT4_I(inode)->i_disksize)
-			ext4_update_i_disksize(inode, new_size);
-	} else {
-		/*
-		* Mark that we allocate beyond EOF so the subsequent truncate
-		* can proceed even if the new size is the same as i_size.
-		*/
-		if ((offset + len) > i_size_read(inode))
-			ext4_set_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
+	if (file->f_flags & O_SYNC && EXT4_SB(inode->i_sb)->s_journal) {
+		ret = jbd2_complete_transaction(EXT4_SB(inode->i_sb)->s_journal,
+						EXT4_I(inode)->i_sync_tid);
 	}
-	ext4_mark_inode_dirty(handle, inode);
-	if (file->f_flags & O_SYNC)
-		ext4_handle_sync(handle);
-
-	ext4_journal_stop(handle);
 out:
 	mutex_unlock(&inode->i_mutex);
 	trace_ext4_fallocate_exit(inode, offset, max_blocks, ret);

fs/ext4/inode.c

@@ -1055,27 +1055,11 @@ static int ext4_write_end(struct file *file,
 	} else
 		copied = block_write_end(file, mapping, pos,
 					 len, copied, page, fsdata);
-
 	/*
-	 * No need to use i_size_read() here, the i_size
-	 * cannot change under us because we hole i_mutex.
-	 *
-	 * But it's important to update i_size while still holding page lock:
+	 * 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);
-		i_size_changed = 1;
-	}
-
-	if (pos + copied > EXT4_I(inode)->i_disksize) {
-		/* We need to mark inode dirty even if
-		 * new_i_size is less that inode->i_size
-		 * but greater than i_disksize. (hint delalloc)
-		 */
-		ext4_update_i_disksize(inode, (pos + copied));
-		i_size_changed = 1;
-	}
+	i_size_changed = ext4_update_inode_size(inode, pos + copied);
 	unlock_page(page);
 	page_cache_release(page);
 
@@ -1123,7 +1107,7 @@ static int ext4_journalled_write_end(struct file *file,
 	int ret = 0, ret2;
 	int partial = 0;
 	unsigned from, to;
-	loff_t new_i_size;
+	int size_changed = 0;
 
 	trace_ext4_journalled_write_end(inode, pos, len, copied);
 	from = pos & (PAGE_CACHE_SIZE - 1);
@@ -1146,20 +1130,18 @@ static int ext4_journalled_write_end(struct file *file,
 		if (!partial)
 			SetPageUptodate(page);
 	}
-	new_i_size = pos + copied;
-	if (new_i_size > inode->i_size)
-		i_size_write(inode, pos+copied);
+	size_changed = ext4_update_inode_size(inode, pos + copied);
 	ext4_set_inode_state(inode, EXT4_STATE_JDATA);
 	EXT4_I(inode)->i_datasync_tid = handle->h_transaction->t_tid;
-	if (new_i_size > EXT4_I(inode)->i_disksize) {
-		ext4_update_i_disksize(inode, new_i_size);
+	unlock_page(page);
+	page_cache_release(page);
+
+	if (size_changed) {
 		ret2 = ext4_mark_inode_dirty(handle, inode);
 		if (!ret)
 			ret = ret2;
 	}
 
-	unlock_page(page);
-	page_cache_release(page);
 	if (pos + len > inode->i_size && ext4_can_truncate(inode))
 		/* if we have allocated more blocks and copied
 		 * less. We will have blocks allocated outside
@@ -2095,6 +2077,7 @@ static int mpage_map_and_submit_extent(handle_t *handle,
 	struct ext4_map_blocks *map = &mpd->map;
 	int err;
 	loff_t disksize;
+	int progress = 0;
 
 	mpd->io_submit.io_end->offset =
 				((loff_t)map->m_lblk) << inode->i_blkbits;
@@ -2111,8 +2094,11 @@ static int mpage_map_and_submit_extent(handle_t *handle,
 			 * is non-zero, a commit should free up blocks.
 			 */
 			if ((err == -ENOMEM) ||
-			    (err == -ENOSPC && ext4_count_free_clusters(sb)))
+			    (err == -ENOSPC && ext4_count_free_clusters(sb))) {
+				if (progress)
+					goto update_disksize;
 				return err;
+			}
 			ext4_msg(sb, KERN_CRIT,
 				 "Delayed block allocation failed for "
 				 "inode %lu at logical offset %llu with"
@@ -2129,15 +2115,17 @@ static int mpage_map_and_submit_extent(handle_t *handle,
 			*give_up_on_write = true;
 			return err;
 		}
+		progress = 1;
 		/*
 		 * Update buffer state, submit mapped pages, and get us new
 		 * extent to map
 		 */
 		err = mpage_map_and_submit_buffers(mpd);
 		if (err < 0)
-			return err;
+			goto update_disksize;
 	} while (map->m_len);
 
+update_disksize:
 	/*
 	 * Update on-disk size after IO is submitted.  Races with
 	 * truncate are avoided by checking i_size under i_data_sem.

fs/ext4/mballoc.c

@@ -1412,6 +1412,8 @@ static void mb_free_blocks(struct inode *inode, struct ext4_buddy *e4b,
 	int last = first + count - 1;
 	struct super_block *sb = e4b->bd_sb;
 
+	if (WARN_ON(count == 0))
+		return;
 	BUG_ON(last >= (sb->s_blocksize << 3));
 	assert_spin_locked(ext4_group_lock_ptr(sb, e4b->bd_group));
 	/* Don't bother if the block group is corrupt. */
@@ -3221,6 +3223,8 @@ static void ext4_discard_allocated_blocks(struct ext4_allocation_context *ac)
 	int err;
 
 	if (pa == NULL) {
+		if (ac->ac_f_ex.fe_len == 0)
+			return;
 		err = ext4_mb_load_buddy(ac->ac_sb, ac->ac_f_ex.fe_group, &e4b);
 		if (err) {
 			/*
@@ -3235,6 +3239,7 @@ static void ext4_discard_allocated_blocks(struct ext4_allocation_context *ac)
 		mb_free_blocks(ac->ac_inode, &e4b, ac->ac_f_ex.fe_start,
 			       ac->ac_f_ex.fe_len);
 		ext4_unlock_group(ac->ac_sb, ac->ac_f_ex.fe_group);
+		ext4_mb_unload_buddy(&e4b);
 		return;
 	}
 	if (pa->pa_type == MB_INODE_PA)

fs/ext4/namei.c

@@ -1227,7 +1227,7 @@ static struct buffer_head * ext4_find_entry (struct inode *dir,
 				   buffer */
 	int num = 0;
 	ext4_lblk_t  nblocks;
-	int i, err;
+	int i, err = 0;
 	int namelen;
 
 	*res_dir = NULL;
@@ -1264,7 +1264,11 @@ static struct buffer_head * ext4_find_entry (struct inode *dir,
 		 * return.  Otherwise, fall back to doing a search the
 		 * old fashioned way.
 		 */
-		if (bh || (err != ERR_BAD_DX_DIR))
+		if (err == -ENOENT)
+			return NULL;
+		if (err && err != ERR_BAD_DX_DIR)
+			return ERR_PTR(err);
+		if (bh)
 			return bh;
 		dxtrace(printk(KERN_DEBUG "ext4_find_entry: dx failed, "
 			       "falling back\n"));
@@ -1295,6 +1299,11 @@ static struct buffer_head * ext4_find_entry (struct inode *dir,
 				}
 				num++;
 				bh = ext4_getblk(NULL, dir, b++, 0, &err);
+				if (unlikely(err)) {
+					if (ra_max == 0)
+						return ERR_PTR(err);
+					break;
+				}
 				bh_use[ra_max] = bh;
 				if (bh)
 					ll_rw_block(READ | REQ_META | REQ_PRIO,
@@ -1417,6 +1426,8 @@ static struct dentry *ext4_lookup(struct inode *dir, struct dentry *dentry, unsi
 		return ERR_PTR(-ENAMETOOLONG);
 
 	bh = ext4_find_entry(dir, &dentry->d_name, &de, NULL);
+	if (IS_ERR(bh))
+		return (struct dentry *) bh;
 	inode = NULL;
 	if (bh) {
 		__u32 ino = le32_to_cpu(de->inode);
@@ -1450,6 +1461,8 @@ struct dentry *ext4_get_parent(struct dentry *child)
 	struct buffer_head *bh;
 
 	bh = ext4_find_entry(child->d_inode, &dotdot, &de, NULL);
+	if (IS_ERR(bh))
+		return (struct dentry *) bh;
 	if (!bh)
 		return ERR_PTR(-ENOENT);
 	ino = le32_to_cpu(de->inode);
@@ -2727,6 +2740,8 @@ static int ext4_rmdir(struct inode *dir, struct dentry *dentry)
 
 	retval = -ENOENT;
 	bh = ext4_find_entry(dir, &dentry->d_name, &de, NULL);
+	if (IS_ERR(bh))
+		return PTR_ERR(bh);
 	if (!bh)
 		goto end_rmdir;
 
@@ -2794,6 +2809,8 @@ static int ext4_unlink(struct inode *dir, struct dentry *dentry)
 
 	retval = -ENOENT;
 	bh = ext4_find_entry(dir, &dentry->d_name, &de, NULL);
+	if (IS_ERR(bh))
+		return PTR_ERR(bh);
 	if (!bh)
 		goto end_unlink;
 
@@ -3121,6 +3138,8 @@ static int ext4_find_delete_entry(handle_t *handle, struct inode *dir,
 	struct ext4_dir_entry_2 *de;
 
 	bh = ext4_find_entry(dir, d_name, &de, NULL);
+	if (IS_ERR(bh))
+		return PTR_ERR(bh);
 	if (bh) {
 		retval = ext4_delete_entry(handle, dir, de, bh);
 		brelse(bh);
@@ -3128,7 +3147,8 @@ static int ext4_find_delete_entry(handle_t *handle, struct inode *dir,
 	return retval;
 }
 
-static void ext4_rename_delete(handle_t *handle, struct ext4_renament *ent)
+static void ext4_rename_delete(handle_t *handle, struct ext4_renament *ent,
+			       int force_reread)
 {
 	int retval;
 	/*
@@ -3140,7 +3160,8 @@ static void ext4_rename_delete(handle_t *handle, struct ext4_renament *ent)
 	if (le32_to_cpu(ent->de->inode) != ent->inode->i_ino ||
 	    ent->de->name_len != ent->dentry->d_name.len ||
 	    strncmp(ent->de->name, ent->dentry->d_name.name,
-		    ent->de->name_len)) {
+		    ent->de->name_len) ||
+	    force_reread) {
 		retval = ext4_find_delete_entry(handle, ent->dir,
 						&ent->dentry->d_name);
 	} else {
@@ -3191,6 +3212,7 @@ static int ext4_rename(struct inode *old_dir, struct dentry *old_dentry,
 		.dentry = new_dentry,
 		.inode = new_dentry->d_inode,
 	};
+	int force_reread;
 	int retval;
 
 	dquot_initialize(old.dir);
@@ -3202,6 +3224,8 @@ static int ext4_rename(struct inode *old_dir, struct dentry *old_dentry,
 		dquot_initialize(new.inode);
 
 	old.bh = ext4_find_entry(old.dir, &old.dentry->d_name, &old.de, NULL);
+	if (IS_ERR(old.bh))
+		return PTR_ERR(old.bh);
 	/*
 	 *  Check for inode number is _not_ due to possible IO errors.
 	 *  We might rmdir the source, keep it as pwd of some process
@@ -3214,6 +3238,10 @@ static int ext4_rename(struct inode *old_dir, struct dentry *old_dentry,
 
 	new.bh = ext4_find_entry(new.dir, &new.dentry->d_name,
 				 &new.de, &new.inlined);
+	if (IS_ERR(new.bh)) {
+		retval = PTR_ERR(new.bh);
+		goto end_rename;
+	}
 	if (new.bh) {
 		if (!new.inode) {
 			brelse(new.bh);
@@ -3246,6 +3274,15 @@ static int ext4_rename(struct inode *old_dir, struct dentry *old_dentry,
 		if (retval)
 			goto end_rename;
 	}
+	/*
+	 * If we're renaming a file within an inline_data dir and adding or
+	 * setting the new dirent causes a conversion from inline_data to
+	 * extents/blockmap, we need to force the dirent delete code to
+	 * re-read the directory, or else we end up trying to delete a dirent
+	 * from what is now the extent tree root (or a block map).
+	 */
+	force_reread = (new.dir->i_ino == old.dir->i_ino &&
+			ext4_test_inode_flag(new.dir, EXT4_INODE_INLINE_DATA));
 	if (!new.bh) {
 		retval = ext4_add_entry(handle, new.dentry, old.inode);
 		if (retval)
@@ -3256,6 +3293,9 @@ static int ext4_rename(struct inode *old_dir, struct dentry *old_dentry,
 		if (retval)
 			goto end_rename;
 	}
+	if (force_reread)
+		force_reread = !ext4_test_inode_flag(new.dir,
+						     EXT4_INODE_INLINE_DATA);
 
 	/*
 	 * Like most other Unix systems, set the ctime for inodes on a
@@ -3267,7 +3307,7 @@ static int ext4_rename(struct inode *old_dir, struct dentry *old_dentry,
 	/*
 	 * ok, that's it
 	 */
-	ext4_rename_delete(handle, &old);
+	ext4_rename_delete(handle, &old, force_reread);
 
 	if (new.inode) {
 		ext4_dec_count(handle, new.inode);
@@ -3330,6 +3370,8 @@ static int ext4_cross_rename(struct inode *old_dir, struct dentry *old_dentry,
 
 	old.bh = ext4_find_entry(old.dir, &old.dentry->d_name,
 				 &old.de, &old.inlined);
+	if (IS_ERR(old.bh))
+		return PTR_ERR(old.bh);
 	/*
 	 *  Check for inode number is _not_ due to possible IO errors.
 	 *  We might rmdir the source, keep it as pwd of some process
@@ -3342,6 +3384,10 @@ static int ext4_cross_rename(struct inode *old_dir, struct dentry *old_dentry,
 
 	new.bh = ext4_find_entry(new.dir, &new.dentry->d_name,
 				 &new.de, &new.inlined);
+	if (IS_ERR(new.bh)) {
+		retval = PTR_ERR(new.bh);
+		goto end_rename;
+	}
 
 	/* RENAME_EXCHANGE case: old *and* new must both exist */
 	if (!new.bh || le32_to_cpu(new.de->inode) != new.inode->i_ino)

fs/ext4/super.c

@@ -3181,9 +3181,9 @@ static int set_journal_csum_feature_set(struct super_block *sb)
 
 	if (EXT4_HAS_RO_COMPAT_FEATURE(sb,
 				       EXT4_FEATURE_RO_COMPAT_METADATA_CSUM)) {
-		/* journal checksum v2 */
+		/* journal checksum v3 */
 		compat = 0;
-		incompat = JBD2_FEATURE_INCOMPAT_CSUM_V2;
+		incompat = JBD2_FEATURE_INCOMPAT_CSUM_V3;
 	} else {
 		/* journal checksum v1 */
 		compat = JBD2_FEATURE_COMPAT_CHECKSUM;
@@ -3205,6 +3205,7 @@ static int set_journal_csum_feature_set(struct super_block *sb)
 		jbd2_journal_clear_features(sbi->s_journal,
 				JBD2_FEATURE_COMPAT_CHECKSUM, 0,
 				JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT |
+				JBD2_FEATURE_INCOMPAT_CSUM_V3 |
 				JBD2_FEATURE_INCOMPAT_CSUM_V2);
 	}
 

fs/jbd2/commit.c

@@ -97,7 +97,7 @@ static void jbd2_commit_block_csum_set(journal_t *j, struct buffer_head *bh)
 	struct commit_header *h;
 	__u32 csum;
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return;
 
 	h = (struct commit_header *)(bh->b_data);
@@ -313,11 +313,11 @@ static __u32 jbd2_checksum_data(__u32 crc32_sum, struct buffer_head *bh)
 	return checksum;
 }
 
-static void write_tag_block(int tag_bytes, journal_block_tag_t *tag,
+static void write_tag_block(journal_t *j, journal_block_tag_t *tag,
 				   unsigned long long block)
 {
 	tag->t_blocknr = cpu_to_be32(block & (u32)~0);
-	if (tag_bytes > JBD2_TAG_SIZE32)
+	if (JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_64BIT))
 		tag->t_blocknr_high = cpu_to_be32((block >> 31) >> 1);
 }
 
@@ -327,7 +327,7 @@ static void jbd2_descr_block_csum_set(journal_t *j,
 	struct jbd2_journal_block_tail *tail;
 	__u32 csum;
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return;
 
 	tail = (struct jbd2_journal_block_tail *)(bh->b_data + j->j_blocksize -
@@ -340,12 +340,13 @@ static void jbd2_descr_block_csum_set(journal_t *j,
 static void jbd2_block_tag_csum_set(journal_t *j, journal_block_tag_t *tag,
 				    struct buffer_head *bh, __u32 sequence)
 {
+	journal_block_tag3_t *tag3 = (journal_block_tag3_t *)tag;
 	struct page *page = bh->b_page;
 	__u8 *addr;
 	__u32 csum32;
 	__be32 seq;
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return;
 
 	seq = cpu_to_be32(sequence);
@@ -355,8 +356,10 @@ static void jbd2_block_tag_csum_set(journal_t *j, journal_block_tag_t *tag,
 			     bh->b_size);
 	kunmap_atomic(addr);
 
-	/* We only have space to store the lower 16 bits of the crc32c. */
-	tag->t_checksum = cpu_to_be16(csum32);
+	if (JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V3))
+		tag3->t_checksum = cpu_to_be32(csum32);
+	else
+		tag->t_checksum = cpu_to_be16(csum32);
 }
 /*
  * jbd2_journal_commit_transaction
@@ -396,7 +399,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 	LIST_HEAD(io_bufs);
 	LIST_HEAD(log_bufs);
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (jbd2_journal_has_csum_v2or3(journal))
 		csum_size = sizeof(struct jbd2_journal_block_tail);
 
 	/*
@@ -690,7 +693,7 @@ void jbd2_journal_commit_transaction(journal_t *journal)
 			tag_flag |= JBD2_FLAG_SAME_UUID;
 
 		tag = (journal_block_tag_t *) tagp;
-		write_tag_block(tag_bytes, tag, jh2bh(jh)->b_blocknr);
+		write_tag_block(journal, tag, jh2bh(jh)->b_blocknr);
 		tag->t_flags = cpu_to_be16(tag_flag);
 		jbd2_block_tag_csum_set(journal, tag, wbuf[bufs],
 					commit_transaction->t_tid);

fs/jbd2/journal.c

@@ -124,7 +124,7 @@ EXPORT_SYMBOL(__jbd2_debug);
 /* Checksumming functions */
 static int jbd2_verify_csum_type(journal_t *j, journal_superblock_t *sb)
 {
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return 1;
 
 	return sb->s_checksum_type == JBD2_CRC32C_CHKSUM;
@@ -145,7 +145,7 @@ static __be32 jbd2_superblock_csum(journal_t *j, journal_superblock_t *sb)
 
 static int jbd2_superblock_csum_verify(journal_t *j, journal_superblock_t *sb)
 {
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return 1;
 
 	return sb->s_checksum == jbd2_superblock_csum(j, sb);
@@ -153,7 +153,7 @@ static int jbd2_superblock_csum_verify(journal_t *j, journal_superblock_t *sb)
 
 static void jbd2_superblock_csum_set(journal_t *j, journal_superblock_t *sb)
 {
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return;
 
 	sb->s_checksum = jbd2_superblock_csum(j, sb);
@@ -1522,21 +1522,29 @@ static int journal_get_superblock(journal_t *journal)
 		goto out;
 	}
 
-	if (JBD2_HAS_COMPAT_FEATURE(journal, JBD2_FEATURE_COMPAT_CHECKSUM) &&
-	    JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2)) {
+	if (jbd2_journal_has_csum_v2or3(journal) &&
+	    JBD2_HAS_COMPAT_FEATURE(journal, JBD2_FEATURE_COMPAT_CHECKSUM)) {
 		/* Can't have checksum v1 and v2 on at the same time! */
 		printk(KERN_ERR "JBD2: Can't enable checksumming v1 and v2 "
 		       "at the same time!\n");
 		goto out;
 	}
 
+	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2) &&
+	    JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V3)) {
+		/* Can't have checksum v2 and v3 at the same time! */
+		printk(KERN_ERR "JBD2: Can't enable checksumming v2 and v3 "
+		       "at the same time!\n");
+		goto out;
+	}
+
 	if (!jbd2_verify_csum_type(journal, sb)) {
 		printk(KERN_ERR "JBD2: Unknown checksum type\n");
 		goto out;
 	}
 
 	/* Load the checksum driver */
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2)) {
+	if (jbd2_journal_has_csum_v2or3(journal)) {
 		journal->j_chksum_driver = crypto_alloc_shash("crc32c", 0, 0);
 		if (IS_ERR(journal->j_chksum_driver)) {
 			printk(KERN_ERR "JBD2: Cannot load crc32c driver.\n");
@@ -1553,7 +1561,7 @@ static int journal_get_superblock(journal_t *journal)
 	}
 
 	/* Precompute checksum seed for all metadata */
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (jbd2_journal_has_csum_v2or3(journal))
 		journal->j_csum_seed = jbd2_chksum(journal, ~0, sb->s_uuid,
 						   sizeof(sb->s_uuid));
 
@@ -1813,8 +1821,14 @@ int jbd2_journal_set_features (journal_t *journal, unsigned long compat,
 	if (!jbd2_journal_check_available_features(journal, compat, ro, incompat))
 		return 0;
 
-	/* Asking for checksumming v2 and v1?  Only give them v2. */
-	if (incompat & JBD2_FEATURE_INCOMPAT_CSUM_V2 &&
+	/* If enabling v2 checksums, turn on v3 instead */
+	if (incompat & JBD2_FEATURE_INCOMPAT_CSUM_V2) {
+		incompat &= ~JBD2_FEATURE_INCOMPAT_CSUM_V2;
+		incompat |= JBD2_FEATURE_INCOMPAT_CSUM_V3;
+	}
+
+	/* Asking for checksumming v3 and v1?  Only give them v3. */
+	if (incompat & JBD2_FEATURE_INCOMPAT_CSUM_V3 &&
 	    compat & JBD2_FEATURE_COMPAT_CHECKSUM)
 		compat &= ~JBD2_FEATURE_COMPAT_CHECKSUM;
 
@@ -1823,8 +1837,8 @@ int jbd2_journal_set_features (journal_t *journal, unsigned long compat,
 
 	sb = journal->j_superblock;
 
-	/* If enabling v2 checksums, update superblock */
-	if (INCOMPAT_FEATURE_ON(JBD2_FEATURE_INCOMPAT_CSUM_V2)) {
+	/* If enabling v3 checksums, update superblock */
+	if (INCOMPAT_FEATURE_ON(JBD2_FEATURE_INCOMPAT_CSUM_V3)) {
 		sb->s_checksum_type = JBD2_CRC32C_CHKSUM;
 		sb->s_feature_compat &=
 			~cpu_to_be32(JBD2_FEATURE_COMPAT_CHECKSUM);
@@ -1842,8 +1856,7 @@ int jbd2_journal_set_features (journal_t *journal, unsigned long compat,
 		}
 
 		/* Precompute checksum seed for all metadata */
-		if (JBD2_HAS_INCOMPAT_FEATURE(journal,
-					      JBD2_FEATURE_INCOMPAT_CSUM_V2))
+		if (jbd2_journal_has_csum_v2or3(journal))
 			journal->j_csum_seed = jbd2_chksum(journal, ~0,
 							   sb->s_uuid,
 							   sizeof(sb->s_uuid));
@@ -1852,7 +1865,8 @@ int jbd2_journal_set_features (journal_t *journal, unsigned long compat,
 	/* If enabling v1 checksums, downgrade superblock */
 	if (COMPAT_FEATURE_ON(JBD2_FEATURE_COMPAT_CHECKSUM))
 		sb->s_feature_incompat &=
-			~cpu_to_be32(JBD2_FEATURE_INCOMPAT_CSUM_V2);
+			~cpu_to_be32(JBD2_FEATURE_INCOMPAT_CSUM_V2 |
+				     JBD2_FEATURE_INCOMPAT_CSUM_V3);
 
 	sb->s_feature_compat    |= cpu_to_be32(compat);
 	sb->s_feature_ro_compat |= cpu_to_be32(ro);
@@ -2165,16 +2179,20 @@ int jbd2_journal_blocks_per_page(struct inode *inode)
  */
 size_t journal_tag_bytes(journal_t *journal)
 {
-	journal_block_tag_t tag;
-	size_t x = 0;
+	size_t sz;
+
+	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V3))
+		return sizeof(journal_block_tag3_t);
+
+	sz = sizeof(journal_block_tag_t);
 
 	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2))
-		x += sizeof(tag.t_checksum);
+		sz += sizeof(__u16);
 
 	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_64BIT))
-		return x + JBD2_TAG_SIZE64;
+		return sz;
 	else
-		return x + JBD2_TAG_SIZE32;
+		return sz - sizeof(__u32);
 }
 
 /*

fs/jbd2/recovery.c

@@ -181,7 +181,7 @@ static int jbd2_descr_block_csum_verify(journal_t *j,
 	__be32 provided;
 	__u32 calculated;
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return 1;
 
 	tail = (struct jbd2_journal_block_tail *)(buf + j->j_blocksize -
@@ -205,7 +205,7 @@ static int count_tags(journal_t *journal, struct buffer_head *bh)
 	int			nr = 0, size = journal->j_blocksize;
 	int			tag_bytes = journal_tag_bytes(journal);
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (jbd2_journal_has_csum_v2or3(journal))
 		size -= sizeof(struct jbd2_journal_block_tail);
 
 	tagp = &bh->b_data[sizeof(journal_header_t)];
@@ -338,10 +338,11 @@ int jbd2_journal_skip_recovery(journal_t *journal)
 	return err;
 }
 
-static inline unsigned long long read_tag_block(int tag_bytes, journal_block_tag_t *tag)
+static inline unsigned long long read_tag_block(journal_t *journal,
+						journal_block_tag_t *tag)
 {
 	unsigned long long block = be32_to_cpu(tag->t_blocknr);
-	if (tag_bytes > JBD2_TAG_SIZE32)
+	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_64BIT))
 		block |= (u64)be32_to_cpu(tag->t_blocknr_high) << 32;
 	return block;
 }
@@ -384,7 +385,7 @@ static int jbd2_commit_block_csum_verify(journal_t *j, void *buf)
 	__be32 provided;
 	__u32 calculated;
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return 1;
 
 	h = buf;
@@ -399,17 +400,21 @@ static int jbd2_commit_block_csum_verify(journal_t *j, void *buf)
 static int jbd2_block_tag_csum_verify(journal_t *j, journal_block_tag_t *tag,
 				      void *buf, __u32 sequence)
 {
+	journal_block_tag3_t *tag3 = (journal_block_tag3_t *)tag;
 	__u32 csum32;
 	__be32 seq;
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return 1;
 
 	seq = cpu_to_be32(sequence);
 	csum32 = jbd2_chksum(j, j->j_csum_seed, (__u8 *)&seq, sizeof(seq));
 	csum32 = jbd2_chksum(j, csum32, buf, j->j_blocksize);
 
-	return tag->t_checksum == cpu_to_be16(csum32);
+	if (JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V3))
+		return tag3->t_checksum == cpu_to_be32(csum32);
+	else
+		return tag->t_checksum == cpu_to_be16(csum32);
 }
 
 static int do_one_pass(journal_t *journal,
@@ -426,6 +431,7 @@ static int do_one_pass(journal_t *journal,
 	int			tag_bytes = journal_tag_bytes(journal);
 	__u32			crc32_sum = ~0; /* Transactional Checksums */
 	int			descr_csum_size = 0;
+	int			block_error = 0;
 
 	/*
 	 * First thing is to establish what we expect to find in the log
@@ -512,8 +518,7 @@ static int do_one_pass(journal_t *journal,
 		switch(blocktype) {
 		case JBD2_DESCRIPTOR_BLOCK:
 			/* Verify checksum first */
-			if (JBD2_HAS_INCOMPAT_FEATURE(journal,
-					JBD2_FEATURE_INCOMPAT_CSUM_V2))
+			if (jbd2_journal_has_csum_v2or3(journal))
 				descr_csum_size =
 					sizeof(struct jbd2_journal_block_tail);
 			if (descr_csum_size > 0 &&
@@ -574,7 +579,7 @@ static int do_one_pass(journal_t *journal,
 					unsigned long long blocknr;
 
 					J_ASSERT(obh != NULL);
-					blocknr = read_tag_block(tag_bytes,
+					blocknr = read_tag_block(journal,
 								 tag);
 
 					/* If the block has been
@@ -598,7 +603,8 @@ static int do_one_pass(journal_t *journal,
 						       "checksum recovering "
 						       "block %llu in log\n",
 						       blocknr);
-						continue;
+						block_error = 1;
+						goto skip_write;
 					}
 
 					/* Find a buffer for the new
@@ -797,7 +803,8 @@ static int do_one_pass(journal_t *journal,
 				success = -EIO;
 		}
 	}
-
+	if (block_error && success == 0)
+		success = -EIO;
 	return success;
 
  failed:
@@ -811,7 +818,7 @@ static int jbd2_revoke_block_csum_verify(journal_t *j,
 	__be32 provided;
 	__u32 calculated;
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return 1;
 
 	tail = (struct jbd2_journal_revoke_tail *)(buf + j->j_blocksize -

fs/jbd2/revoke.c

@@ -91,8 +91,8 @@
 #include <linux/list.h>
 #include <linux/init.h>
 #include <linux/bio.h>
-#endif
 #include <linux/log2.h>
+#endif
 
 static struct kmem_cache *jbd2_revoke_record_cache;
 static struct kmem_cache *jbd2_revoke_table_cache;
@@ -597,7 +597,7 @@ static void write_one_revoke_record(journal_t *journal,
 	offset = *offsetp;
 
 	/* Do we need to leave space at the end for a checksum? */
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (jbd2_journal_has_csum_v2or3(journal))
 		csum_size = sizeof(struct jbd2_journal_revoke_tail);
 
 	/* Make sure we have a descriptor with space left for the record */
@@ -644,7 +644,7 @@ static void jbd2_revoke_csum_set(journal_t *j, struct buffer_head *bh)
 	struct jbd2_journal_revoke_tail *tail;
 	__u32 csum;
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (!jbd2_journal_has_csum_v2or3(j))
 		return;
 
 	tail = (struct jbd2_journal_revoke_tail *)(bh->b_data + j->j_blocksize -

include/linux/jbd2.h

@@ -159,7 +159,11 @@ typedef struct journal_header_s
  * journal_block_tag (in the descriptor).  The other h_chksum* fields are
  * not used.
  *
- * Checksum v1 and v2 are mutually exclusive features.
+ * If FEATURE_INCOMPAT_CSUM_V3 is set, the descriptor block uses
+ * journal_block_tag3_t to store a full 32-bit checksum.  Everything else
+ * is the same as v2.
+ *
+ * Checksum v1, v2, and v3 are mutually exclusive features.
  */
 struct commit_header {
 	__be32		h_magic;
@@ -179,6 +183,14 @@ struct commit_header {
  * raw struct shouldn't be used for pointer math or sizeof() - use
  * journal_tag_bytes(journal) instead to compute this.
  */
+typedef struct journal_block_tag3_s
+{
+	__be32		t_blocknr;	/* The on-disk block number */
+	__be32		t_flags;	/* See below */
+	__be32		t_blocknr_high; /* most-significant high 32bits. */
+	__be32		t_checksum;	/* crc32c(uuid+seq+block) */
+} journal_block_tag3_t;
+
 typedef struct journal_block_tag_s
 {
 	__be32		t_blocknr;	/* The on-disk block number */
@@ -187,9 +199,6 @@ typedef struct journal_block_tag_s
 	__be32		t_blocknr_high; /* most-significant high 32bits. */
 } journal_block_tag_t;
 
-#define JBD2_TAG_SIZE32 (offsetof(journal_block_tag_t, t_blocknr_high))
-#define JBD2_TAG_SIZE64 (sizeof(journal_block_tag_t))
-
 /* Tail of descriptor block, for checksumming */
 struct jbd2_journal_block_tail {
 	__be32		t_checksum;	/* crc32c(uuid+descr_block) */
@@ -284,6 +293,7 @@ typedef struct journal_superblock_s
 #define JBD2_FEATURE_INCOMPAT_64BIT		0x00000002
 #define JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT	0x00000004
 #define JBD2_FEATURE_INCOMPAT_CSUM_V2		0x00000008
+#define JBD2_FEATURE_INCOMPAT_CSUM_V3		0x00000010
 
 /* Features known to this kernel version: */
 #define JBD2_KNOWN_COMPAT_FEATURES	JBD2_FEATURE_COMPAT_CHECKSUM
@@ -291,7 +301,8 @@ typedef struct journal_superblock_s
 #define JBD2_KNOWN_INCOMPAT_FEATURES	(JBD2_FEATURE_INCOMPAT_REVOKE | \
 					JBD2_FEATURE_INCOMPAT_64BIT | \
 					JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT | \
-					JBD2_FEATURE_INCOMPAT_CSUM_V2)
+					JBD2_FEATURE_INCOMPAT_CSUM_V2 | \
+					JBD2_FEATURE_INCOMPAT_CSUM_V3)
 
 #ifdef __KERNEL__
 
@@ -1296,6 +1307,15 @@ static inline int tid_geq(tid_t x, tid_t y)
 extern int jbd2_journal_blocks_per_page(struct inode *inode);
 extern size_t journal_tag_bytes(journal_t *journal);
 
+static inline int jbd2_journal_has_csum_v2or3(journal_t *journal)
+{
+	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2) ||
+	    JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V3))
+		return 1;
+
+	return 0;
+}
+
 /*
  * We reserve t_outstanding_credits >> JBD2_CONTROL_BLOCKS_SHIFT for
  * transaction control blocks.