Charles Hedrick

Suppose this is accepted. There's currently a command nfs4_setacl in Linux. will that now set an NFS4 acl across nfs? That command doesn't work locally. Ideally there would be a...

I believe with Unix the typical approach would be

The comment about nfs4_getfacl not working leads to questions. Just how does this patch fit into the broader system? As I understand it, the NFS4 protocol supports these ACLs. Does...

Further issue: We use ACLs to set default permissions. We set them from an NFS client using POSIX if the system is mounted NFS3 or NFS4 if it is mounted...

Can you explain this new XDR encoding? As far as I can tell from online documents, the NFS 4.1 protocol defines a format for transmitting ACLs over the wire, and...

By the way, for a `client` to use the NFS4.1 protocol to set ACLs, it's not just system.nfs4_acl. There are also attributes system.nfs4_sacl and system.nfs4_dacl. The protocol provide NFS attributes...

> > Generally patching the linux kernel is required in order for proper interoperability / compatibility with the linux kernel VFS. There are permissions concepts (such as the rather complex...

Sort of. You can set an nfs4 ACL by setting a zfs specific xattr. There is a python progrsm to do that. It goes over nfs, but as a generic...

Without the kernel patch, I believe default ACLs will break. To work correctly they need create permissions and umaxk to be separate. Since the kernel will think the file system...

Let me give you the specifics. The kernel calls to create a file and directory include a permission. Typically a file is created with 666. That gets combined with the...