Re: [PATCH v5 02/14] ntfs: update in-memory, on-disk structures and headers

[Christoph Hellwig]

On Tue, Jan 20, 2026 at 01:27:55PM +0900, Namjae Jeon wrote:
> On Mon, Jan 19, 2026 at 4:05 PM Christoph Hellwig <hch@xxxxxx> wrote:
> >
> > On Sun, Jan 18, 2026 at 01:54:06PM +0900, Namjae Jeon wrote:
> > > > It seem like big_ntfs_inode is literally only used in the conversion
> > > > helpers below.  Are there are a lot of these "extent inode" so that
> > > > not having the vfs inode for them is an actual saving?
> > > Right, In NTFS, a base MFT record (represented by the base ntfs_inode)
> > > requires a struct inode to interact with the VFS. However, a single
> > > file can have multiple extent MFT records to store additional
> > > attributes. These extent inodes are managed internally by the base
> > > inode and do not need to be visible to the VFS.
> >
> > What are typical numbers of the extra extent inodes?  If they are rare,
> > you might be able to simplify the code a bit by just always allocating
> > the vfs_inode even if it's not really used.
> Regarding the typical numbers, in most cases, It will require zero or
> only a few extra extent inodes. Okay, I will move vfs_inode to
> ntfs_inode.

This was just thinking out loud.  If it doesn't help to significantly
simplify things, don't bother.