Re: [RFC][PATCH 16/42] now we can fold open_check_o_direct() into do_dentry_open()

[Linus Torvalds]

I like the patch, I hate the commit message.

It makes sense right now in this sequence, but I'd really like the
commit message to say _why_ this sequence led up to this point.

Right now I still remember you trying this, and having to revert it
because it didn't work before all the fput/put_filp issues. But a year
from now? Five years from now?

So at least a "now that fput() works regardless of how far the open
got.." kind of explanation, ok?

              Linus