Messages in this thread | | | From | Hannes Frederic Sowa <> | Subject | Re: use-after-free in sock_wake_async | Date | Thu, 26 Nov 2015 18:15:49 +0100 |
| |
On Thu, Nov 26, 2015, at 18:09, Eric Dumazet wrote: > On Thu, 2015-11-26 at 18:03 +0100, Hannes Frederic Sowa wrote: > > > My rationale was like this: we already have rcu to free the wq, so we > > don't add any more callbacks as current code. sock_alloc is right now > > 1136 bytes, which is huge, like 18 cachelines. I wouldn't think it does > > matter a lot as we thrash anyway. tcp_sock is like 45 cachelines right > > now, hui. > > Again, these speculations are really killing me. > > Can you wait I send my patches ?
Eric, I wasn't targeting upstream. I just had this idea and wanted to show you that it is possible. I didn't send an official patch and wouldn't do that without proper performance analysis.
> They work just great, and there is no risk of perf regression with added > RCU freeing of inodes :(
I agree. Your patch is certainly the best thing for net while this complete rcuification should never ever hit net.
> It is Thanks Giving in the US, vacation time.
Enjoy the time! :)
Bye, Hannes
| |