/* SPDX-License-Identifier: GPL-2.0+ *//* * RCU expedited grace periods * * Copyright IBM Corporation, 2016 * * Authors: Paul E. McKenney <paulmck@linux.ibm.com> */ #include <linux/console.h>#include <linux/lockdep.h> static void rcu_exp_handler(void *unused);static int rcu_print_task_exp_stall(struct rcu_node *rnp);static void rcu_exp_print_detail_task_stall_rnp(struct rcu_node *rnp); /* * Record the start of an expedited grace period. */static void rcu_exp_gp_seq_start(void){ rcu_seq_start(&rcu_state.expedited_sequence); rcu_poll_gp_seq_start_unlocked(&rcu_state.gp_seq_polled_exp_snap);} /* * Return the value that the expedited-grace-period counter will have * at the end of the current grace period. */static __maybe_unused unsigned long rcu_exp_gp_seq_endval(void){ return rcu_seq_endval(&rcu_state.expedited_sequence);} /* * Record the end of an expedited grace period. */static void rcu_exp_gp_seq_end(void){ rcu_poll_gp_seq_end_unlocked(&rcu_state.gp_seq_polled_exp_snap); rcu_seq_end(&rcu_state.expedited_sequence); smp_mb(); /* Ensure that consecutive grace periods serialize. */} /* * Take a snapshot of the expedited-grace-period counter, which is the * earliest value that will indicate that a full grace period has * elapsed since the current time. */static unsigned long rcu_exp_gp_seq_snap(void){ unsigned long s; smp_mb(); /* Caller's modifications seen first by other CPUs. */ s = rcu_seq_snap(&rcu_state.expedited_sequence); trace_rcu_exp_grace_period(rcu_state.name, s, TPS("snap")); return s;} /* * Given a counter snapshot from rcu_exp_gp_seq_snap(), return true * if a full expedited grace period has elapsed since that snapshot * was taken. */static bool rcu_exp_gp_seq_done(unsigned long s){ return rcu_seq_done(&rcu_state.expedited_sequence, s);} /* * Reset the ->expmaskinit values in the rcu_node tree to reflect any * recent CPU-online activity. Note that these masks are not cleared * when CPUs go offline, so they reflect the union of all CPUs that have * ever been online. This means that this function normally takes its * no-work-to-do fastpath. */static void sync_exp_reset_tree_hotplug(void){ bool done; unsigned long flags; unsigned long mask; unsigned long oldmask; int ncpus = smp_load_acquire(&rcu_state.ncpus); /* Order vs. locking. */ struct rcu_node *rnp; struct rcu_node *rnp_up; /* If no new CPUs onlined since last time, nothing to do. */ if (likely(ncpus == rcu_state.ncpus_snap)) return; rcu_state.ncpus_snap = ncpus; /* * Each pass through the following loop propagates newly onlined * CPUs for the current rcu_node structure up the rcu_node tree. */ rcu_for_each_leaf_node(rnp) { raw_spin_lock_irqsave_rcu_node(rnp, flags); if (rnp->expmaskinit == rnp->expmaskinitnext) { raw_spin_unlock_irqrestore_rcu_node(rnp, flags); continue; /* No new CPUs, nothing to do. */ } /* Update this node's mask, track old value for propagation. */ oldmask = rnp->expmaskinit; rnp->expmaskinit = rnp->expmaskinitnext; raw_spin_unlock_irqrestore_rcu_node(rnp, flags); /* If was already nonzero, nothing to propagate. */ if (oldmask) continue; /* Propagate the new CPU up the tree. */ mask = rnp->grpmask; rnp_up = rnp->parent; done = false; while (rnp_up) { raw_spin_lock_irqsave_rcu_node(rnp_up, flags); if (rnp_up->expmaskinit) done = true; rnp_up->expmaskinit |= mask; raw_spin_unlock_irqrestore_rcu_node(rnp_up, flags); if (done) break; mask = rnp_up->grpmask; rnp_up = rnp_up->parent; } }} /* * Reset the ->expmask values in the rcu_node tree in preparation for * a new expedited grace period. */static void __maybe_unused sync_exp_reset_tree(void){ unsigned long flags; struct rcu_node *rnp; sync_exp_reset_tree_hotplug(); rcu_for_each_node_breadth_first(rnp) { raw_spin_lock_irqsave_rcu_node(rnp, flags); WARN_ON_ONCE(rnp->expmask); WRITE_ONCE(rnp->expmask, rnp->expmaskinit); /* * Need to wait for any blocked tasks as well. Note that * additional blocking tasks will also block the expedited GP * until such time as the ->expmask bits are cleared. */ if (rcu_is_leaf_node(rnp) && rcu_preempt_has_tasks(rnp)) WRITE_ONCE(rnp->exp_tasks, rnp->blkd_tasks.next); raw_spin_unlock_irqrestore_rcu_node(rnp, flags); }} /* * Return non-zero if there is no RCU expedited grace period in progress * for the specified rcu_node structure, in other words, if all CPUs and * tasks covered by the specified rcu_node structure have done their bit * for the current expedited grace period. */static bool sync_rcu_exp_done(struct rcu_node *rnp){ raw_lockdep_assert_held_rcu_node(rnp); return READ_ONCE(rnp->exp_tasks) == NULL && READ_ONCE(rnp->expmask) == 0;} /* * Like sync_rcu_exp_done(), but where the caller does not hold the * rcu_node's ->lock. */static bool sync_rcu_exp_done_unlocked(struct rcu_node *rnp){ unsigned long flags; bool ret; raw_spin_lock_irqsave_rcu_node(rnp, flags); ret = sync_rcu_exp_done(rnp); raw_spin_unlock_irqrestore_rcu_node(rnp, flags); return ret;} /* * Report the exit from RCU read-side critical section for the last task * that queued itself during or before the current expedited preemptible-RCU * grace period. This event is reported either to the rcu_node structure on * which the task was queued or to one of that rcu_node structure's ancestors, * recursively up the tree. (Calm down, calm down, we do the recursion * iteratively!) */static void __rcu_report_exp_rnp(struct rcu_node *rnp, bool wake, unsigned long flags) __releases(rnp->lock){ unsigned long mask; raw_lockdep_assert_held_rcu_node(rnp); for (;;) { if (!sync_rcu_exp_done(rnp)) { if (!rnp->expmask) rcu_initiate_boost(rnp, flags); else raw_spin_unlock_irqrestore_rcu_node(rnp, flags); break; } if (rnp->parent == NULL) { raw_spin_unlock_irqrestore_rcu_node(rnp, flags); if (wake) swake_up_one(&rcu_state.expedited_wq); break; } mask = rnp->grpmask; raw_spin_unlock_rcu_node(rnp); /* irqs remain disabled */ rnp = rnp->parent; raw_spin_lock_rcu_node(rnp); /* irqs already disabled */ WARN_ON_ONCE(!(rnp->expmask & mask)); WRITE_ONCE(rnp->expmask, rnp->expmask & ~mask); }} /* * Report expedited quiescent state for specified node. This is a * lock-acquisition wrapper function for __rcu_report_exp_rnp(). */static void __maybe_unused rcu_report_exp_rnp(struct rcu_node *rnp, bool wake){ unsigned long flags; raw_spin_lock_irqsave_rcu_node(rnp, flags); __rcu_report_exp_rnp(rnp, wake, flags);} /* * Report expedited quiescent state for multiple CPUs, all covered by the * specified leaf rcu_node structure, which is acquired by the caller. */static void rcu_report_exp_cpu_mult(struct rcu_node *rnp, unsigned long flags, unsigned long mask_in, bool wake) __releases(rnp->lock){ int cpu; unsigned long mask; struct rcu_data *rdp; raw_lockdep_assert_held_rcu_node(rnp); if (!(rnp->expmask & mask_in)) { raw_spin_unlock_irqrestore_rcu_node(rnp, flags); return; } mask = mask_in & rnp->expmask; WRITE_ONCE(rnp->expmask, rnp->expmask & ~mask); for_each_leaf_node_cpu_mask(rnp, cpu, mask) { rdp = per_cpu_ptr(&rcu_data, cpu); if (!IS_ENABLED(CONFIG_NO_HZ_FULL) || !rdp->rcu_forced_tick_exp) continue; rdp->rcu_forced_tick_exp = false; tick_dep_clear_cpu(cpu, TICK_DEP_BIT_RCU_EXP); } __rcu_report_exp_rnp(rnp, wake, flags); /* Releases rnp->lock. */} /* * Report expedited quiescent state for specified rcu_data (CPU). */static void rcu_report_exp_rdp(struct rcu_data *rdp){ unsigned long flags; struct rcu_node *rnp = rdp->mynode; raw_spin_lock_irqsave_rcu_node(rnp, flags); WRITE_ONCE(rdp->cpu_no_qs.b.exp, false); ASSERT_EXCLUSIVE_WRITER(rdp->cpu_no_qs.b.exp); rcu_report_exp_cpu_mult(rnp, flags, rdp->grpmask, true);} /* Common code for work-done checking. */static bool sync_exp_work_done(unsigned long s){ if (rcu_exp_gp_seq_done(s)) { trace_rcu_exp_grace_period(rcu_state.name, s, TPS("done")); /* * Order GP completion with preceding accesses. Order also GP * completion with post GP update side accesses. Pairs with * rcu_seq_end(). */ smp_mb(); return true; } return false;} /* * Funnel-lock acquisition for expedited grace periods. Returns true * if some other task completed an expedited grace period that this task * can piggy-back on, and with no mutex held. Otherwise, returns false * with the mutex held, indicating that the caller must actually do the * expedited grace period. */static bool exp_funnel_lock(unsigned long s){ struct rcu_data *rdp = per_cpu_ptr(&rcu_data, raw_smp_processor_id()); struct rcu_node *rnp = rdp->mynode; struct rcu_node *rnp_root = rcu_get_root(); /* Low-contention fastpath. */ if (ULONG_CMP_LT(READ_ONCE(rnp->exp_seq_rq), s) && (rnp == rnp_root || ULONG_CMP_LT(READ_ONCE(rnp_root->exp_seq_rq), s)) && mutex_trylock(&rcu_state.exp_mutex)) goto fastpath; /* * Each pass through the following loop works its way up * the rcu_node tree, returning if others have done the work or * otherwise falls through to acquire ->exp_mutex. The mapping * from CPU to rcu_node structure can be inexact, as it is just * promoting locality and is not strictly needed for correctness. */ for (; rnp != NULL; rnp = rnp->parent) { if (sync_exp_work_done(s)) return true; /* Work not done, either wait here or go up. */ spin_lock(&rnp->exp_lock); if (ULONG_CMP_GE(rnp->exp_seq_rq, s)) { /* Someone else doing GP, so wait for them. */ spin_unlock(&rnp->exp_lock); trace_rcu_exp_funnel_lock(rcu_state.name, rnp->level, rnp->grplo, rnp->grphi, TPS("wait")); wait_event(rnp->exp_wq[rcu_seq_ctr(s) & 0x3], sync_exp_work_done(s)); return true; } WRITE_ONCE(rnp->exp_seq_rq, s); /* Followers can wait on us. */ spin_unlock(&rnp->exp_lock); trace_rcu_exp_funnel_lock(rcu_state.name, rnp->level, rnp->grplo, rnp->grphi, TPS("nxtlvl")); } mutex_lock(&rcu_state.exp_mutex);fastpath: if (sync_exp_work_done(s)) { mutex_unlock(&rcu_state.exp_mutex); return true; } rcu_exp_gp_seq_start(); trace_rcu_exp_grace_period(rcu_state.name, s, TPS("start")); return false;} /* * Select the CPUs within the specified rcu_node that the upcoming * expedited grace period needs to wait for. */static void __sync_rcu_exp_select_node_cpus(struct rcu_exp_work *rewp){ int cpu; unsigned long flags; unsigned long mask_ofl_test; unsigned long mask_ofl_ipi; int ret; struct rcu_node *rnp = container_of(rewp, struct rcu_node, rew); raw_spin_lock_irqsave_rcu_node(rnp, flags); /* Each pass checks a CPU for identity, offline, and idle. */ mask_ofl_test = 0; for_each_leaf_node_cpu_mask(rnp, cpu, rnp->expmask) { struct rcu_data *rdp = per_cpu_ptr(&rcu_data, cpu); unsigned long mask = rdp->grpmask; int snap; if (raw_smp_processor_id() == cpu || !(rnp->qsmaskinitnext & mask)) { mask_ofl_test |= mask; } else { /* * Full ordering between remote CPU's post idle accesses * and updater's accesses prior to current GP (and also * the started GP sequence number) is enforced by * rcu_seq_start() implicit barrier, relayed by kworkers * locking and even further by smp_mb__after_unlock_lock() * barriers chained all the way throughout the rnp locking * tree since sync_exp_reset_tree() and up to the current * leaf rnp locking. * * Ordering between remote CPU's pre idle accesses and * post grace period updater's accesses is enforced by the * below acquire semantic. */ snap = ct_rcu_watching_cpu_acquire(cpu); if (rcu_watching_snap_in_eqs(snap)) mask_ofl_test |= mask; else rdp->exp_watching_snap = snap; } } mask_ofl_ipi = rnp->expmask & ~mask_ofl_test; raw_spin_unlock_irqrestore_rcu_node(rnp, flags); /* IPI the remaining CPUs for expedited quiescent state. */ for_each_leaf_node_cpu_mask(rnp, cpu, mask_ofl_ipi) { struct rcu_data *rdp = per_cpu_ptr(&rcu_data, cpu); unsigned long mask = rdp->grpmask; retry_ipi: if (rcu_watching_snap_stopped_since(rdp, rdp->exp_watching_snap)) { mask_ofl_test |= mask; continue; } if (get_cpu() == cpu) { mask_ofl_test |= mask; put_cpu(); continue; } ret = smp_call_function_single(cpu, rcu_exp_handler, NULL, 0); put_cpu(); /* The CPU will report the QS in response to the IPI. */ if (!ret) continue; /* Failed, raced with CPU hotplug operation. */ raw_spin_lock_irqsave_rcu_node(rnp, flags); if ((rnp->qsmaskinitnext & mask) && (rnp->expmask & mask)) { /* Online, so delay for a bit and try again. */ raw_spin_unlock_irqrestore_rcu_node(rnp, flags); trace_rcu_exp_grace_period(rcu_state.name, rcu_exp_gp_seq_endval(), TPS("selectofl")); schedule_timeout_idle(1); goto retry_ipi; } /* CPU really is offline, so we must report its QS. */ if (rnp->expmask & mask) mask_ofl_test |= mask; raw_spin_unlock_irqrestore_rcu_node(rnp, flags); } /* Report quiescent states for those that went offline. */ if (mask_ofl_test) { raw_spin_lock_irqsave_rcu_node(rnp, flags); rcu_report_exp_cpu_mult(rnp, flags, mask_ofl_test, false); }} static void rcu_exp_sel_wait_wake(unsigned long s); static void sync_rcu_exp_select_node_cpus(struct kthread_work *wp){ struct rcu_exp_work *rewp = container_of(wp, struct rcu_exp_work, rew_work); __sync_rcu_exp_select_node_cpus(rewp);} static inline bool rcu_exp_worker_started(void){ return !!READ_ONCE(rcu_exp_gp_kworker);} static inline bool rcu_exp_par_worker_started(struct rcu_node *rnp){ return !!READ_ONCE(rnp->exp_kworker);} static inline void sync_rcu_exp_select_cpus_queue_work(struct rcu_node *rnp){ kthread_init_work(&rnp->rew.rew_work, sync_rcu_exp_select_node_cpus); /* * Use rcu_exp_par_gp_kworker, because flushing a work item from * another work item on the same kthread worker can result in * deadlock. */ kthread_queue_work(READ_ONCE(rnp->exp_kworker), &rnp->rew.rew_work);} static inline void sync_rcu_exp_select_cpus_flush_work(struct rcu_node *rnp){ kthread_flush_work(&rnp->rew.rew_work);} /* * Work-queue handler to drive an expedited grace period forward. */static void wait_rcu_exp_gp(struct kthread_work *wp){ struct rcu_exp_work *rewp; rewp = container_of(wp, struct rcu_exp_work, rew_work); rcu_exp_sel_wait_wake(rewp->rew_s);} static inline void synchronize_rcu_expedited_queue_work(struct rcu_exp_work *rew){ kthread_init_work(&rew->rew_work, wait_rcu_exp_gp); kthread_queue_work(rcu_exp_gp_kworker, &rew->rew_work);} Lines 1–500 of 1123. Load more
/* SPDX-License-Identifier: GPL-2.0+ */ /* * RCU expedited grace periods * * Copyright IBM Corporation, 2016 * * Authors: Paul E. McKenney <paulmck@linux.ibm.com> */ #include <linux/console.h> #include <linux/lockdep.h> static void rcu_exp_handler(void *unused); static int rcu_print_task_exp_stall(struct rcu_node *rnp); static void rcu_exp_print_detail_task_stall_rnp(struct rcu_node *rnp); /* * Record the start of an expedited grace period. */ static void rcu_exp_gp_seq_start(void) { rcu_seq_start(&rcu_state.expedited_sequence); rcu_poll_gp_seq_start_unlocked(&rcu_state.gp_seq_polled_exp_snap); } /* * Return the value that the expedited-grace-period counter will have * at the end of the current grace period. */ static __maybe_unused unsigned long rcu_exp_gp_seq_endval(void) { return rcu_seq_endval(&rcu_state.expedited_sequence); } /* * Record the end of an expedited grace period. */ static void rcu_exp_gp_seq_end(void) { rcu_poll_gp_seq_end_unlocked(&rcu_state.gp_seq_polled_exp_snap); rcu_seq_end(&rcu_state.expedited_sequence); smp_mb(); /* Ensure that consecutive grace periods serialize. */ } /* * Take a snapshot of the expedited-grace-period counter, which is the * earliest value that will indicate that a full grace period has * elapsed since the current time. */ static unsigned long rcu_exp_gp_seq_snap(void) { unsigned long s; smp_mb(); /* Caller's modifications seen first by other CPUs. */ s = rcu_seq_snap(&rcu_state.expedited_sequence); trace_rcu_exp_grace_period(rcu_state.name, s, TPS("snap")); return s; } /* * Given a counter snapshot from rcu_exp_gp_seq_snap(), return true * if a full expedited grace period has elapsed since that snapshot * was taken. */ static bool rcu_exp_gp_seq_done(unsigned long s) { return rcu_seq_done(&rcu_state.expedited_sequence, s); } /* * Reset the ->expmaskinit values in the rcu_node tree to reflect any * recent CPU-online activity. Note that these masks are not cleared * when CPUs go offline, so they reflect the union of all CPUs that have * ever been online. This means that this function normally takes its * no-work-to-do fastpath. */ static void sync_exp_reset_tree_hotplug(void) { bool done; unsigned long flags; unsigned long mask; unsigned long oldmask; int ncpus = smp_load_acquire(&rcu_state.ncpus); /* Order vs. locking. */ struct rcu_node *rnp; struct rcu_node *rnp_up; /* If no new CPUs onlined since last time, nothing to do. */ if (likely(ncpus == rcu_state.ncpus_snap)) return; rcu_state.ncpus_snap = ncpus; /* * Each pass through the following loop propagates newly onlined * CPUs for the current rcu_node structure up the rcu_node tree. */ rcu_for_each_leaf_node(rnp) { raw_spin_lock_irqsave_rcu_node(rnp, flags); if (rnp->expmaskinit == rnp->expmaskinitnext) { raw_spin_unlock_irqrestore_rcu_node(rnp, flags); continue; /* No new CPUs, nothing to do. */ } /* Update this node's mask, track old value for propagation. */ oldmask = rnp->expmaskinit; rnp->expmaskinit = rnp->expmaskinitnext; raw_spin_unlock_irqrestore_rcu_node(rnp, flags); /* If was already nonzero, nothing to propagate. */ if (oldmask) continue; /* Propagate the new CPU up the tree. */ mask = rnp->grpmask; rnp_up = rnp->parent; done = false; while (rnp_up) { raw_spin_lock_irqsave_rcu_node(rnp_up, flags); if (rnp_up->expmaskinit) done = true; rnp_up->expmaskinit |= mask; raw_spin_unlock_irqrestore_rcu_node(rnp_up, flags); if (done) break; mask = rnp_up->grpmask; rnp_up = rnp_up->parent; } } } /* * Reset the ->expmask values in the rcu_node tree in preparation for * a new expedited grace period. */ static void __maybe_unused sync_exp_reset_tree(void) { unsigned long flags; struct rcu_node *rnp; sync_exp_reset_tree_hotplug(); rcu_for_each_node_breadth_first(rnp) { raw_spin_lock_irqsave_rcu_node(rnp, flags); WARN_ON_ONCE(rnp->expmask); WRITE_ONCE(rnp->expmask, rnp->expmaskinit); /* * Need to wait for any blocked tasks as well. Note that * additional blocking tasks will also block the expedited GP * until such time as the ->expmask bits are cleared. */ if (rcu_is_leaf_node(rnp) && rcu_preempt_has_tasks(rnp)) WRITE_ONCE(rnp->exp_tasks, rnp->blkd_tasks.next); raw_spin_unlock_irqrestore_rcu_node(rnp, flags); } } /* * Return non-zero if there is no RCU expedited grace period in progress * for the specified rcu_node structure, in other words, if all CPUs and * tasks covered by the specified rcu_node structure have done their bit * for the current expedited grace period. */ static bool sync_rcu_exp_done(struct rcu_node *rnp) { raw_lockdep_assert_held_rcu_node(rnp); return READ_ONCE(rnp->exp_tasks) == NULL && READ_ONCE(rnp->expmask) == 0; } /* * Like sync_rcu_exp_done(), but where the caller does not hold the * rcu_node's ->lock. */ static bool sync_rcu_exp_done_unlocked(struct rcu_node *rnp) { unsigned long flags; bool ret; raw_spin_lock_irqsave_rcu_node(rnp, flags); ret = sync_rcu_exp_done(rnp); raw_spin_unlock_irqrestore_rcu_node(rnp, flags); return ret; } /* * Report the exit from RCU read-side critical section for the last task * that queued itself during or before the current expedited preemptible-RCU * grace period. This event is reported either to the rcu_node structure on * which the task was queued or to one of that rcu_node structure's ancestors, * recursively up the tree. (Calm down, calm down, we do the recursion * iteratively!) */ static void __rcu_report_exp_rnp(struct rcu_node *rnp, bool wake, unsigned long flags) __releases(rnp->lock) { unsigned long mask; raw_lockdep_assert_held_rcu_node(rnp); for (;;) { if (!sync_rcu_exp_done(rnp)) { if (!rnp->expmask) rcu_initiate_boost(rnp, flags); else raw_spin_unlock_irqrestore_rcu_node(rnp, flags); break; } if (rnp->parent == NULL) { raw_spin_unlock_irqrestore_rcu_node(rnp, flags); if (wake) swake_up_one(&rcu_state.expedited_wq); break; } mask = rnp->grpmask; raw_spin_unlock_rcu_node(rnp); /* irqs remain disabled */ rnp = rnp->parent; raw_spin_lock_rcu_node(rnp); /* irqs already disabled */ WARN_ON_ONCE(!(rnp->expmask & mask)); WRITE_ONCE(rnp->expmask, rnp->expmask & ~mask); } } /* * Report expedited quiescent state for specified node. This is a * lock-acquisition wrapper function for __rcu_report_exp_rnp(). */ static void __maybe_unused rcu_report_exp_rnp(struct rcu_node *rnp, bool wake) { unsigned long flags; raw_spin_lock_irqsave_rcu_node(rnp, flags); __rcu_report_exp_rnp(rnp, wake, flags); } /* * Report expedited quiescent state for multiple CPUs, all covered by the * specified leaf rcu_node structure, which is acquired by the caller. */ static void rcu_report_exp_cpu_mult(struct rcu_node *rnp, unsigned long flags, unsigned long mask_in, bool wake) __releases(rnp->lock) { int cpu; unsigned long mask; struct rcu_data *rdp; raw_lockdep_assert_held_rcu_node(rnp); if (!(rnp->expmask & mask_in)) { raw_spin_unlock_irqrestore_rcu_node(rnp, flags); return; } mask = mask_in & rnp->expmask; WRITE_ONCE(rnp->expmask, rnp->expmask & ~mask); for_each_leaf_node_cpu_mask(rnp, cpu, mask) { rdp = per_cpu_ptr(&rcu_data, cpu); if (!IS_ENABLED(CONFIG_NO_HZ_FULL) || !rdp->rcu_forced_tick_exp) continue; rdp->rcu_forced_tick_exp = false; tick_dep_clear_cpu(cpu, TICK_DEP_BIT_RCU_EXP); } __rcu_report_exp_rnp(rnp, wake, flags); /* Releases rnp->lock. */ } /* * Report expedited quiescent state for specified rcu_data (CPU). */ static void rcu_report_exp_rdp(struct rcu_data *rdp) { unsigned long flags; struct rcu_node *rnp = rdp->mynode; raw_spin_lock_irqsave_rcu_node(rnp, flags); WRITE_ONCE(rdp->cpu_no_qs.b.exp, false); ASSERT_EXCLUSIVE_WRITER(rdp->cpu_no_qs.b.exp); rcu_report_exp_cpu_mult(rnp, flags, rdp->grpmask, true); } /* Common code for work-done checking. */ static bool sync_exp_work_done(unsigned long s) { if (rcu_exp_gp_seq_done(s)) { trace_rcu_exp_grace_period(rcu_state.name, s, TPS("done")); /* * Order GP completion with preceding accesses. Order also GP * completion with post GP update side accesses. Pairs with * rcu_seq_end(). */ smp_mb(); return true; } return false; } /* * Funnel-lock acquisition for expedited grace periods. Returns true * if some other task completed an expedited grace period that this task * can piggy-back on, and with no mutex held. Otherwise, returns false * with the mutex held, indicating that the caller must actually do the * expedited grace period. */ static bool exp_funnel_lock(unsigned long s) { struct rcu_data *rdp = per_cpu_ptr(&rcu_data, raw_smp_processor_id()); struct rcu_node *rnp = rdp->mynode; struct rcu_node *rnp_root = rcu_get_root(); /* Low-contention fastpath. */ if (ULONG_CMP_LT(READ_ONCE(rnp->exp_seq_rq), s) && (rnp == rnp_root || ULONG_CMP_LT(READ_ONCE(rnp_root->exp_seq_rq), s)) && mutex_trylock(&rcu_state.exp_mutex)) goto fastpath; /* * Each pass through the following loop works its way up * the rcu_node tree, returning if others have done the work or * otherwise falls through to acquire ->exp_mutex. The mapping * from CPU to rcu_node structure can be inexact, as it is just * promoting locality and is not strictly needed for correctness. */ for (; rnp != NULL; rnp = rnp->parent) { if (sync_exp_work_done(s)) return true; /* Work not done, either wait here or go up. */ spin_lock(&rnp->exp_lock); if (ULONG_CMP_GE(rnp->exp_seq_rq, s)) { /* Someone else doing GP, so wait for them. */ spin_unlock(&rnp->exp_lock); trace_rcu_exp_funnel_lock(rcu_state.name, rnp->level, rnp->grplo, rnp->grphi, TPS("wait")); wait_event(rnp->exp_wq[rcu_seq_ctr(s) & 0x3], sync_exp_work_done(s)); return true; } WRITE_ONCE(rnp->exp_seq_rq, s); /* Followers can wait on us. */ spin_unlock(&rnp->exp_lock); trace_rcu_exp_funnel_lock(rcu_state.name, rnp->level, rnp->grplo, rnp->grphi, TPS("nxtlvl")); } mutex_lock(&rcu_state.exp_mutex); fastpath: if (sync_exp_work_done(s)) { mutex_unlock(&rcu_state.exp_mutex); return true; } rcu_exp_gp_seq_start(); trace_rcu_exp_grace_period(rcu_state.name, s, TPS("start")); return false; } /* * Select the CPUs within the specified rcu_node that the upcoming * expedited grace period needs to wait for. */ static void __sync_rcu_exp_select_node_cpus(struct rcu_exp_work *rewp) { int cpu; unsigned long flags; unsigned long mask_ofl_test; unsigned long mask_ofl_ipi; int ret; struct rcu_node *rnp = container_of(rewp, struct rcu_node, rew); raw_spin_lock_irqsave_rcu_node(rnp, flags); /* Each pass checks a CPU for identity, offline, and idle. */ mask_ofl_test = 0; for_each_leaf_node_cpu_mask(rnp, cpu, rnp->expmask) { struct rcu_data *rdp = per_cpu_ptr(&rcu_data, cpu); unsigned long mask = rdp->grpmask; int snap; if (raw_smp_processor_id() == cpu || !(rnp->qsmaskinitnext & mask)) { mask_ofl_test |= mask; } else { /* * Full ordering between remote CPU's post idle accesses * and updater's accesses prior to current GP (and also * the started GP sequence number) is enforced by * rcu_seq_start() implicit barrier, relayed by kworkers * locking and even further by smp_mb__after_unlock_lock() * barriers chained all the way throughout the rnp locking * tree since sync_exp_reset_tree() and up to the current * leaf rnp locking. * * Ordering between remote CPU's pre idle accesses and * post grace period updater's accesses is enforced by the * below acquire semantic. */ snap = ct_rcu_watching_cpu_acquire(cpu); if (rcu_watching_snap_in_eqs(snap)) mask_ofl_test |= mask; else rdp->exp_watching_snap = snap; } } mask_ofl_ipi = rnp->expmask & ~mask_ofl_test; raw_spin_unlock_irqrestore_rcu_node(rnp, flags); /* IPI the remaining CPUs for expedited quiescent state. */ for_each_leaf_node_cpu_mask(rnp, cpu, mask_ofl_ipi) { struct rcu_data *rdp = per_cpu_ptr(&rcu_data, cpu); unsigned long mask = rdp->grpmask; retry_ipi: if (rcu_watching_snap_stopped_since(rdp, rdp->exp_watching_snap)) { mask_ofl_test |= mask; continue; } if (get_cpu() == cpu) { mask_ofl_test |= mask; put_cpu(); continue; } ret = smp_call_function_single(cpu, rcu_exp_handler, NULL, 0); put_cpu(); /* The CPU will report the QS in response to the IPI. */ if (!ret) continue; /* Failed, raced with CPU hotplug operation. */ raw_spin_lock_irqsave_rcu_node(rnp, flags); if ((rnp->qsmaskinitnext & mask) && (rnp->expmask & mask)) { /* Online, so delay for a bit and try again. */ raw_spin_unlock_irqrestore_rcu_node(rnp, flags); trace_rcu_exp_grace_period(rcu_state.name, rcu_exp_gp_seq_endval(), TPS("selectofl")); schedule_timeout_idle(1); goto retry_ipi; } /* CPU really is offline, so we must report its QS. */ if (rnp->expmask & mask) mask_ofl_test |= mask; raw_spin_unlock_irqrestore_rcu_node(rnp, flags); } /* Report quiescent states for those that went offline. */ if (mask_ofl_test) { raw_spin_lock_irqsave_rcu_node(rnp, flags); rcu_report_exp_cpu_mult(rnp, flags, mask_ofl_test, false); } } static void rcu_exp_sel_wait_wake(unsigned long s); static void sync_rcu_exp_select_node_cpus(struct kthread_work *wp) { struct rcu_exp_work *rewp = container_of(wp, struct rcu_exp_work, rew_work); __sync_rcu_exp_select_node_cpus(rewp); } static inline bool rcu_exp_worker_started(void) { return !!READ_ONCE(rcu_exp_gp_kworker); } static inline bool rcu_exp_par_worker_started(struct rcu_node *rnp) { return !!READ_ONCE(rnp->exp_kworker); } static inline void sync_rcu_exp_select_cpus_queue_work(struct rcu_node *rnp) { kthread_init_work(&rnp->rew.rew_work, sync_rcu_exp_select_node_cpus); /* * Use rcu_exp_par_gp_kworker, because flushing a work item from * another work item on the same kthread worker can result in * deadlock. */ kthread_queue_work(READ_ONCE(rnp->exp_kworker), &rnp->rew.rew_work); } static inline void sync_rcu_exp_select_cpus_flush_work(struct rcu_node *rnp) { kthread_flush_work(&rnp->rew.rew_work); } /* * Work-queue handler to drive an expedited grace period forward. */ static void wait_rcu_exp_gp(struct kthread_work *wp) { struct rcu_exp_work *rewp; rewp = container_of(wp, struct rcu_exp_work, rew_work); rcu_exp_sel_wait_wake(rewp->rew_s); } static inline void synchronize_rcu_expedited_queue_work(struct rcu_exp_work *rew) { kthread_init_work(&rew->rew_work, wait_rcu_exp_gp); kthread_queue_work(rcu_exp_gp_kworker, &rew->rew_work); }