Messages in this thread | | | Date | Tue, 29 Apr 2014 15:33:54 -0400 | From | Benjamin LaHaise <> | Subject | Re: [PATCH 0/1] aio: change exit_aio() to load mm->ioctx_table once and avoid rcu_read_lock() |
| |
On Tue, Apr 29, 2014 at 08:39:15PM +0200, Oleg Nesterov wrote: > Completely untested and I know nothing about aio, thus needs an ack > from maintainers or should be ignored. > > But exit_aio() looks "obviously unnecessarily overcomplicated", I was > really puzzled when I looked at this code by accident. > > Kent, could you also explain kioctx->dead is atomic_t? This looks > pointless? afaics atomic_ buys nothing in this case.
If it wasn't atomic, it would need to be volatile.
-ben -- "Thought is the essence of where you are now."
| |