Messages in this thread | | | Date | Tue, 24 Mar 2015 16:48:55 +0100 | From | Peter Zijlstra <> | Subject | Re: [RFC][PATCH 0/4] Another go at ktime_get_raw_fast_ns() |
| |
On Tue, Mar 24, 2015 at 08:20:07AM -0700, John Stultz wrote: > On Thu, Mar 19, 2015 at 2:31 AM, Peter Zijlstra <peterz@infradead.org> wrote: > > Hi John, > > > > So yeah, I just didn't try hard enough it seems; how about this? > > Yea, this looks much better! I'll queue it up for testing and if that > goes well include it for 4.1.
I have a few patches that depend on it, so if you're ok with it I'd like to carry the lot. But yes please test ;-)
> > Also, I think we should double check that everything passed to > > tk_setup_internals(.clock) is either static storage or RCU-sched freed, > > because with the whole latch thing one CPUs (fast) access might still be > > in progress while the other is freeing the clock struct. > > > > Not sure what the lifetime rules of struct clocksource are here. > > At one point it was install and never remove, but I think Thomas > changed that awhile back, so I think your concern here is a good one.
OK, I'll try and dig around a bit on this.
| |