Re: [RFC PATCH v1 1/7] fs: Add inode_get_ino() and implement get_ino() for NFS

[Christoph Hellwig <hch@infradead.org>]

On Thu, Oct 17, 2024 at 06:43:38PM +0200, Jan Kara wrote:
> On Thu 17-10-24 08:25:19, Christoph Hellwig wrote:
> > On Thu, Oct 17, 2024 at 11:15:49AM -0400, Paul Moore wrote:
> > > Also good to know, thanks.  However, at this point the lack of a clear
> > > answer is making me wonder a bit more about inode numbers in the view
> > > of VFS developers; do you folks care about inode numbers?
> > 
> > The VFS itself does not care much about inode numbers.  The Posix API
> > does, although btrfs ignores that and seems to get away with that
> > (mostly because applications put in btrfs-specific hacks).
> 
> Well, btrfs plays tricks with *device* numbers, right? Exactly so that
> st_ino + st_dev actually stay unique for each file. Whether it matters for
> audit I don't dare to say :). Bcachefs does not care and returns non-unique
> inode numbers.

But st_ino + st_dev is the only thing Posix and thus historically Linux
has guaranteed to applications.  So if st_dev is unique, but you need
an unknown scope in which it is unique it might as well not be for
that purpose.  And I think for any kind of audit report that is true
as well.