On Mon, 2022-05-09 at 17:42 -0400, Vivek Goyal wrote: > On Mon, May 09, 2022 at 05:08:56PM +0200, Christian Brauner wrote: > > [..] > > Having "xattr" in the system call name is just confusing. These are > > fundamentally not "real" xattrs and we shouldn't mix semantics. > > There > > should be a clear distinction between traditional xattrs and this > > vfs > > and potentially fs information providing interface. > > > > Just thinking about what the manpage would look like. We would need > > to > > add a paragraph to xattr(7) explaining that in addition to the > > system.*, > > security.*, user.* and other namespaces we now also have a set of > > namespaces that function as ways to get information about mounts or > > other things instead of information attached to specific inodes. > > > > That's super random imho. If I were to be presented with this > > manpage > > I'd wonder if someone was too lazy to add a proper new system call > > with > > it's own semantics for this and just stuffed it into an existing > > API > > because it provided matching system call arguments. We can add a > > new > > system call. It's not that we're running out of them. > > FWIW, I also felt that using xattr API to get some sort of mount info > felt > very non-intutive. Yeah, people looking for this function simply wouldn't know to look here ... Ian
