Re: [Patch] VFS : mount lock scalability for files systems withoutmount point (WAS vfsmount lock issues on very large ppc64 box)

From: Christoph Hellwig
Date: Sat Jul 23 2011 - 09:24:40 EST

I think you actually want this done in kern_mount_data, as both
ipc and proc want long-term references as well. I also suspect with
additional creep of container awareness more internal mounts will switch
to kern_mount_data. Al, what do you think about simply passing the
private data argument to kern_mount and kill kern_mount_data? It's not
like the additional argument is going to cause us any pain.

> +struct vfsmount *kern_mount(struct file_system_type *type)
> +{
> + struct vfsmount *mnt;
> +
> + mnt = kern_mount_data(type, NULL);
> + if (!IS_ERR(mnt)) {
> + /* it is a longterm mount, don't release mnt until */
> + /* we unmount before file sys is unregistered */

Please use normal kernel comment style, e.g.

* This is a longterm mount, don't release mnt until we umount
* it just before unregister_filesystem().

Adding proper kerneldoc comments for the kern_mount/umount function that
explain things in more detail would also be nice.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at