Messages in this thread | | | Date | Thu, 23 Aug 2012 12:54:56 -0500 | From | Nathan Zimmer <> | Subject | Re: [PATCH] fs/proc: Move kfree outside pde_unload_lock |
| |
On 08/22/2012 04:42 PM, Eric Dumazet wrote: > On Wed, 2012-08-22 at 20:28 +0200, Eric Dumazet wrote: > >> Thats interesting, but if you really want this to fly, one RCU >> conversion would be much better ;) >> >> pde_users would be an atomic_t and you would avoid the spinlock >> contention. > Here is what I had in mind, I would be interested to know how it helps a 512 core machine ;) > Thanks, I knew if I just took my time and read the rcu documentation thoroughly that the answer would be forthcoming. ;)
Unfortunately I have to wait till tomorrow to get big box and test it.
| |