Re: [PATCH DRAFT RFC 34/34] buffer: port block device access to files and get rid of bd_inode access

[Jan Kara <jack@suse.cz>]

On Mon 08-01-24 16:52:02, Dave Chinner wrote:
> On Wed, Jan 03, 2024 at 01:55:32PM +0100, Christian Brauner wrote:
> > Signed-off-by: Christian Brauner <brauner@kernel.org>
> > ---
> >  block/fops.c                  |  1 +
> >  drivers/md/md-bitmap.c        |  1 +
> >  fs/affs/file.c                |  1 +
> >  fs/btrfs/inode.c              |  1 +
> >  fs/buffer.c                   | 69 ++++++++++++++++++++++---------------------
> >  fs/direct-io.c                |  2 +-
> >  fs/erofs/data.c               |  7 +++--
> >  fs/erofs/internal.h           |  1 +
> >  fs/erofs/zmap.c               |  1 +
> >  fs/ext2/inode.c               |  8 +++--
> >  fs/ext4/inode.c               |  6 ++--
> >  fs/ext4/super.c               |  6 ++--
> >  fs/f2fs/data.c                |  6 +++-
> >  fs/f2fs/f2fs.h                |  1 +
> >  fs/fuse/dax.c                 |  1 +
> >  fs/gfs2/aops.c                |  1 +
> >  fs/gfs2/bmap.c                |  1 +
> >  fs/hpfs/file.c                |  1 +
> >  fs/jbd2/commit.c              |  1 +
> >  fs/jbd2/journal.c             | 26 +++++++++-------
> >  fs/jbd2/recovery.c            |  6 ++--
> >  fs/jbd2/revoke.c              | 10 +++----
> >  fs/jbd2/transaction.c         |  1 +
> >  fs/mpage.c                    |  5 +++-
> >  fs/nilfs2/btnode.c            |  2 ++
> >  fs/nilfs2/gcinode.c           |  1 +
> >  fs/nilfs2/mdt.c               |  1 +
> >  fs/nilfs2/page.c              |  2 ++
> >  fs/nilfs2/recovery.c          | 20 ++++++-------
> >  fs/nilfs2/the_nilfs.c         |  1 +
> >  fs/ntfs/aops.c                |  3 ++
> >  fs/ntfs/file.c                |  1 +
> >  fs/ntfs/mft.c                 |  2 ++
> >  fs/ntfs3/fsntfs.c             |  8 ++---
> >  fs/ntfs3/inode.c              |  1 +
> >  fs/ntfs3/super.c              |  2 +-
> >  fs/ocfs2/journal.c            |  2 +-
> >  fs/reiserfs/journal.c         |  8 ++---
> >  fs/reiserfs/reiserfs.h        |  6 ++--
> >  fs/reiserfs/tail_conversion.c |  1 +
> >  fs/xfs/xfs_iomap.c            |  7 +++--
> >  fs/zonefs/file.c              |  2 ++
> >  include/linux/buffer_head.h   | 45 +++++++++++++++-------------
> >  include/linux/iomap.h         |  1 +
> >  include/linux/jbd2.h          |  6 ++--
> >  45 files changed, 172 insertions(+), 114 deletions(-)
> > 
> > diff --git a/block/fops.c b/block/fops.c
> > index e831196dafac..6557b71c7657 100644
> > --- a/block/fops.c
> > +++ b/block/fops.c
> > @@ -381,6 +381,7 @@ static int blkdev_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> >  	loff_t isize = i_size_read(inode);
> >  
> >  	iomap->bdev = bdev;
> > +	BUG_ON(true /* TODO(brauner): This is the only place where we don't go from inode->i_sb->s_f_bdev for obvious reasons. Thoughts? */);
> 
> Maybe block devices should have their own struct file created when the
> block device is instantiated and torn down when the block device is
> trashed?

OK, but is there a problem with I_BDEV() which is currently used in
blkdev_iomap_begin()?

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR