Re: [PATCH 14/39] ovl: stack file ops

[Al Viro]

On Tue, Jun 12, 2018 at 03:29:26AM +0100, Al Viro wrote:

> It might (or might not) work for the filesystems you'd been testing
> on, but it's a lot of trouble waiting to happen.  Hell, try and use
> ecryptfs as lower layer, see how fast it'll blow up.  Sure, it's
> a dumb testcase, but I don't see how to check if something more
> realistic is trouble-free.
> 
> I'd been trying to come up with some way to salvage that kludge of yours,
> but I don't see any solutions.  We don't have good proxies for "this
> filesystem might be unsafe as lower layer" ;-/

Note that anything that uses file_dentry() anywhere near ->open(),
->read_iter() or ->write_iter() is an instant trouble with your scheme.
Such as
int nfs_open(struct inode *inode, struct file *filp)
{
        struct nfs_open_context *ctx;

        ctx = alloc_nfs_open_context(file_dentry(filp), filp->f_mode, filp);
        if (IS_ERR(ctx)) 
                return PTR_ERR(ctx);
        nfs_file_set_open_context(filp, ctx);
        put_nfs_open_context(ctx);
        nfs_fscache_open_file(inode, filp);
        return 0;
}

You do want to support NFS for lower layers, right?