ext4: clean up ext4_remove_blocks() return

Message ID 20170713151703.GA6383@localhost.localdomain
State Rejected
Headers show

Commit Message

Eric Whitney July 13, 2017, 3:17 p.m.
ext4_remove_blocks() never returns anything other than 0, signifying
success.  Convert the function to void type to make this immediately
obvious at the call site.

Signed-off-by: Eric Whitney <enwlinux@gmail.com>
---
 fs/ext4/extents.c | 14 +++++---------
 1 file changed, 5 insertions(+), 9 deletions(-)

Comments

Theodore Ts'o July 13, 2017, 10:33 p.m. | #1
On Thu, Jul 13, 2017 at 11:17:03AM -0400, Eric Whitney wrote:
> ext4_remove_blocks() never returns anything other than 0, signifying
> success.  Convert the function to void type to make this immediately
> obvious at the call site.
> 
> Signed-off-by: Eric Whitney <enwlinux@gmail.com>

Actually, what we should do is to bubble errors up from
ext4_remove_blocks to the callers of ext4_ext_rm_leaf().  At the
moment it doesn't return any errors, but eventually we should be
returning errors all the way up to ext4 truncate.

The reason why it's a little tricky is it's more than just not
deleting the call to ext4_std_error() and then returning error up to
the caller.  We need to make sure that file system is in a
self-consistent state before we return an error up to the caller.

Cheers,

					- Ted
Eric Whitney July 14, 2017, 9:01 p.m. | #2
* Theodore Ts'o <tytso@mit.edu>:
> On Thu, Jul 13, 2017 at 11:17:03AM -0400, Eric Whitney wrote:
> > ext4_remove_blocks() never returns anything other than 0, signifying
> > success.  Convert the function to void type to make this immediately
> > obvious at the call site.
> > 
> > Signed-off-by: Eric Whitney <enwlinux@gmail.com>
> 
> Actually, what we should do is to bubble errors up from
> ext4_remove_blocks to the callers of ext4_ext_rm_leaf().  At the
> moment it doesn't return any errors, but eventually we should be
> returning errors all the way up to ext4 truncate.
> 
> The reason why it's a little tricky is it's more than just not
> deleting the call to ext4_std_error() and then returning error up to
> the caller.  We need to make sure that file system is in a
> self-consistent state before we return an error up to the caller.
> 
> Cheers,
> 
> 					- Ted
>

OK, that confirms a suspicion I had.  That looks more like a small project
than a quick cleanup, so I'll withdraw the patch and perhaps revisit this
in the future.

Thanks,
Eric

Patch

diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index e0a8425..27da180 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -2496,10 +2496,10 @@  static inline int get_default_free_blocks_flags(struct inode *inode)
 	return 0;
 }
 
-static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
-			      struct ext4_extent *ex,
-			      long long *partial_cluster,
-			      ext4_lblk_t from, ext4_lblk_t to)
+static void ext4_remove_blocks(handle_t *handle, struct inode *inode,
+			       struct ext4_extent *ex,
+			       long long *partial_cluster,
+			       ext4_lblk_t from, ext4_lblk_t to)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
 	unsigned short ee_len = ext4_ext_get_actual_len(ex);
@@ -2592,7 +2592,6 @@  static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
 		ext4_error(sbi->s_sb, "strange request: removal(2) "
 			   "%u-%u from %u:%u",
 			   from, to, le32_to_cpu(ex->ee_block), ee_len);
-	return 0;
 }
 
 
@@ -2720,10 +2719,7 @@  ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
 		if (err)
 			goto out;
 
-		err = ext4_remove_blocks(handle, inode, ex, partial_cluster,
-					 a, b);
-		if (err)
-			goto out;
+		ext4_remove_blocks(handle, inode, ex, partial_cluster, a, b);
 
 		if (num == 0)
 			/* this extent is removed; mark slot entirely unused */