Le Wed, Dec 20, 2023 at 04:24:35PM +0800, Z qiang a écrit : > > > > When the CPU goes idle for the last time during the CPU down hotplug > > process, RCU reports a final quiescent state for the current CPU. If > > this quiescent state propagates up to the top, some tasks may then be > > woken up to complete the grace period: the main grace period kthread > > and/or the expedited main workqueue (or kworker). > > > > If those kthreads have a SCHED_FIFO policy, the wake up can indirectly > > arm the RT bandwith timer to the local offline CPU. Since this happens > > after hrtimers have been migrated at CPUHP_AP_HRTIMERS_DYING stage, the > > timer gets ignored. Therefore if the RCU kthreads are waiting for RT > > bandwidth to be available, they may never be actually scheduled. > > > > In the rcutree_report_cpu_dead(), the rcuog kthreads may also be wakeup in > do_nocb_deferred_wakeup(), if the rcuog kthreads is rt-fifo and wakeup happen, > the rt_period_active is set 1 and enqueue hrtimer to offline CPU in > do_start_rt_bandwidth(), > after that, we invoke swake_up_one_online() send ipi to online CPU, due to the > rt_period_active is 1, the rt-bandwith hrtimer will not enqueue to online CPU. > any thoughts? Duh, you're right, that one too. How many more? This hrtimer situation is scary... Thanks.
