On Sat, Mar 26, 2016 at 05:09:55PM -0400, Theodore Ts'o wrote:

I have the following patches in the ext4.git tree which I plan to push
to Linus as bug fixes during this development cycles.  Al, are you
happy with Miklos's v2 version of "fs: add file_dentrY()" patch?

I'm not really happy, but I guess we'll have to live with that approach.
I would still like to point out that *any* use of file_dentry() (or
file->f_path.dentry, for that matter) is a serious red flag - odds are,
the code using it is broken, possibly by design.

I still don't understand the locking in ext4 crypto, and I'm not at all
convinced that it is correct ;-/  OTOH, there are filesystems where we
really need dentry and (hopefully) treat it sanely enough, so consider
that helper and method ACKed.  I can take it via vfs.git, or leave it
to ext4.git, or do some combination thereof (e.g. the infrastructure
goes into never-rebased branch in vfs.git, with ext4/btrfs/nfs merging
from it).  Up to you...

Al, still bloody unhappy about the whole pile of worms...