Hello! This series reduces memory footprint (RCU callbacks posted by no-CBs CPUs) by providing separate rcuog kthreads that wait for grace periods, thus avoiding the current situation in which callbacks are delayed while the leader rcuo kthreads invokes callbacks: 1. Rename rcu_data fields to prepare for forward-progress work. 2. Update comments to prepare for forward-progress work. 3. Provide separate no-CBs grace-period kthreads. 4. Rename nocb_follower_wait() to nocb_cb_wait(). 5. Rename wake_nocb_leader() to wake_nocb_gp(). 6. Rename __wake_nocb_leader() to __wake_nocb_gp(). 7. Rename wake_nocb_leader_defer() to wake_nocb_gp_defer(). 8. Rename rcu_organize_nocb_kthreads() local variable. 9. Rename and document no-CB CB kthread sleep trace event. 10. Rename rcu_nocb_leader_stride kernel boot parameter. 11. Print gp/cb kthread hierarchy if dump_tree. Thanx, Paul ------------------------------------------------------------------------ Documentation/admin-guide/kernel-parameters.txt | 13 - include/trace/events/rcu.h | 3 kernel/rcu/tree.h | 28 +- kernel/rcu/tree_plugin.h | 312 ++++++++++++------------ 4 files changed, 183 insertions(+), 173 deletions(-)
