The first patch should only be moving code.
To be able to access cpu_rq in sched.c:
raw_spin_lock_irq(&cpu_rq(cpu)->lock);
void unlock_runqueue(unsigned int cpu)
wrapper is added and kernel/sched.h created.
The locks and sched.h are removed again in the second
patch, this should be the only "real" change.
There atomic64_t is introduced in a per_cpu_ptr variable
to avoid the lock.
Not sure about allocating atomic64_t like that is
correct. Review is appreciated.
Both patches are compile tested on various archs and configs.
Thomas
_______________________________________________
Containers mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/containers
_______________________________________________
Devel mailing list
[email protected]
https://openvz.org/mailman/listinfo/devel
