Messages in this thread | | | Date | Wed, 15 Jan 2014 11:30:32 +0100 | From | Miklos Szeredi <> | Subject | Re: [PATCH 04/11] vfs: add renameat2 syscall |
| |
On Wed, Jan 15, 2014 at 07:11:16AM +0900, Tetsuo Handa wrote: > Miklos Szeredi wrote: > > + > > + if (flags) > > + return -EOPNOTSUPP; > > + > > If (at least for now) only ext4 interprets renameat2() flags, > I think adding a new member to "struct inode_operations" and > check it like > > if (!old_dir->i_op->swapname) > return -EOPNOTSUPP; > > or
This is not just about swapname, but other rename variants too (non-overwriting rename is also included in the patchset).
> > if (!(old_dir->i_op->supported_features & flags)) > return -EOPNOTSUPP;
Or rather old_dir->i_sb->s_type->fs_flags. We access it due to FS_RENAME_DOES_D_MOVE anyway.
I like this variant, because now filesystems need to explicitly add a flag to *enable* the functionality and not a check to disable it.
Objections?
Thanks, Miklos
| |