Messages in this thread | | | Date | Thu, 16 Oct 2014 15:54:42 +0200 | Subject | Re: [PATCH 0/5] fuse: handle release synchronously (v4) | From | Linus Torvalds <> |
| |
On Thu, Oct 16, 2014 at 3:43 PM, Miklos Szeredi <miklos@szeredi.hu> wrote: > > One idea is to change ->flush() so it's responsible for fput()-ing the > file. That way we could take control of the actual refcount > decrement. There are only 20 flush instances in the tree, so it > wouldn't be a huge change.
Since that *still* wouldn't fix the problem with the whole "count elevated by other things" issue, I really don't want to hear about these random broken hacks that are fundamentally broken crap.
Really. Stop cc'ing me with "let's implement this hack that cannot work in general". I'm not interested. There's a reason we don't do this. We don't make up random hacks that we know cannot work in the general case.
Linus
| |