Messages in this thread | | | Subject | Re: [PATCH RT 3/9][RFC] [PATCH 3/9] lglock/rt: Use non-rt for_each_cpu() in -rt code | From | Steven Rostedt <> | Date | Fri, 02 Mar 2012 09:20:48 -0500 |
| |
On Fri, 2012-03-02 at 12:55 +0530, Srivatsa S. Bhat wrote: > On 03/02/2012 12:25 AM, Steven Rostedt wrote: > } \ > > + register_hotcpu_notifier(&name##_lg_cpu_notifier); \ > > + get_online_cpus(); \ > > + for_each_online_cpu(i) \ > > + cpu_set(i, name##_cpus); \ > > > This can be further improved. We don't really need this loop. We can replace > it with: > > cpumask_copy(&name##_cpus, cpu_online_mask); > > (as pointed out by Ingo. See: https://lkml.org/lkml/2012/2/29/93 and > https://lkml.org/lkml/2012/2/29/153). > > I will try sending a patch for this to non-RT after the numerous patches > currently flying around this code (in non-RT) settle down.. > >
Yeah, I thought that was funny too, but I wanted RT to be close to mainline. My thoughts were to fix the broken hotplug code at the time and I wasn't thinking too much on improving mainline ;-)
-- Steve
| |