Re: a question about clone_mnt in namespace.c

[Al Viro <viro@ZenIV.linux.org.uk>]

On Wed, Jun 20, 2012 at 10:54:00AM +0800, yushang wrote:
> Hi dear all,
> I have a question about the mnt_master logic in clone_mnt , look at
> following lines:
> ...
> } else if (!(flag & CL_PRIVATE)) {
>   if ((flag & CL_MAKE_SHARED) || IS_MNT_SHARED(old))
> 	list_add(&mnt->mnt_share, &old->mnt_share);
>   if (IS_MNT_SLAVE(old))
> 	list_add(&mnt->mnt_slave, &old->mnt_slave);
>     mnt->mnt_master = old->mnt_master;
> }
> ...
> I think the following lines are equivalent, right ? or I've missed
> something ? many thanks!
> ...
> } else if (!(flag & CL_PRIVATE)) {
>   if ((flag & CL_MAKE_SHARED) || IS_MNT_SHARED(old))
> 	list_add(&mnt->mnt_share, &old->mnt_share);
>   if (IS_MNT_SLAVE(old)) {
> 	list_add(&mnt->mnt_slave, &old->mnt_slave);
>         mnt->mnt_master = old->mnt_master;
>   }
> }

Yes; we are guaranteed that mnt->master is NULL after alloc_vfsmnt(),
so that would be an equivalent transformation.  Matter of taste,
mostly - performance win is negligible and IMO your variant is
slightly less explicit.