Messages in this thread | | | Date | Sun, 14 Dec 2014 12:23:40 -0800 | From | "Paul E. McKenney" <> | Subject | Re: [PATCH 1/3] tracepoints: Do not use call_rcu_sched() before early_initcall() |
| |
On Sun, Dec 14, 2014 at 01:25:56PM -0500, Steven Rostedt wrote: > On Sun, 14 Dec 2014 10:18:35 -0800 > "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> wrote: > > > > With your feed back, and because I would like this to go into 3.19, I > > > would like to keep the current patch as is (with the read_mostly > > > update, which I'm currently testing). We can always change it later > > > after call_rcu() has been changed. > > > > Completely agreed! ;-) > > For this late in the game, we need to play it safe. > > I got rid of my tracepoint patch and moved trace_init() to after > rcu_init(), which I think is good enough for Thomas. Thomas? > > I'll start testing that now.
That should work as well.
> Later, if you could add a rcu_init() version that lets us call > call_rcu_sched() just after mm_init() then we could move trace_init() > up a bit more.
OK, will put something together for that. Good to hear that this is not needed indefinitely early -- that would be a bit harder. ;-)
Thanx, Paul
| |