On Fri, Jan 10, 2014 at 8:06 AM, Al Viro [off-list ref] wrote:

Sure, we can separate the simple "just do call_rcu(...->free_inode)" case
and hit it whenever full ->free_inode is there and ->destroy_inode isn't.
Not too pretty, but removal of tons of boilerplate might be worth doing
that anyway.

Yeah.

But ->destroy_inode() is still needed for cases where fs
has its own idea of inode lifetime rules.  Again, check what XFS is doing
in that area...

Ok, so we can't change destroy_inode, and we'd need to add a new op
for just freeing it.

Painful mainly because there are so many filesystems, but it shouldn't
be *complicated*.

There's an extra source of headache, BTW - what about the "LSM stacking"
crowd and their plans?

LSM stacking is a pipedream right  now anyway, isn't it? It's been
talked about for years and years, I've never seen a patch-set that is
even remotely something we'd seriously consider.

            Linus