#include <linux/wait.h>
#include <linux/kthread.h>
#include <linux/prefetch.h>
+#include <linux/delay.h>
+#include <linux/stop_machine.h>
#include "rcutree.h"
#include <trace/events/rcu.h>
module_param(qhimark, int, 0);
module_param(qlowmark, int, 0);
-int rcu_cpu_stall_suppress __read_mostly;
+int rcu_cpu_stall_suppress __read_mostly; /* 1 = suppress stall warnings. */
+int rcu_cpu_stall_timeout __read_mostly = CONFIG_RCU_CPU_STALL_TIMEOUT;
+
module_param(rcu_cpu_stall_suppress, int, 0644);
+module_param(rcu_cpu_stall_timeout, int, 0644);
static void force_quiescent_state(struct rcu_state *rsp, int relaxed);
static int rcu_pending(int cpu);
static int rcu_implicit_offline_qs(struct rcu_data *rdp)
{
/*
- * If the CPU is offline, it is in a quiescent state. We can
- * trust its state not to change because interrupts are disabled.
+ * If the CPU is offline for more than a jiffy, it is in a quiescent
+ * state. We can trust its state not to change because interrupts
+ * are disabled. The reason for the jiffy's worth of slack is to
+ * handle CPUs initializing on the way up and finding their way
+ * to the idle loop on the way down.
*/
- if (cpu_is_offline(rdp->cpu)) {
+ if (cpu_is_offline(rdp->cpu) &&
+ ULONG_CMP_LT(rdp->rsp->gp_start + 2, jiffies)) {
trace_rcu_fqs(rdp->rsp->name, rdp->gpnum, rdp->cpu, "ofl");
rdp->offline_fqs++;
return 1;
}
-
- /*
- * The CPU is online, so send it a reschedule IPI. This forces
- * it through the scheduler, and (inefficiently) also handles cases
- * where idle loops fail to inform RCU about the CPU being idle.
- */
- if (rdp->cpu != smp_processor_id())
- smp_send_reschedule(rdp->cpu);
- else
- set_need_resched();
- rdp->resched_ipi++;
return 0;
}
}
EXPORT_SYMBOL(rcu_is_cpu_idle);
+#ifdef CONFIG_HOTPLUG_CPU
+
+/*
+ * Is the current CPU online? Disable preemption to avoid false positives
+ * that could otherwise happen due to the current CPU number being sampled,
+ * this task being preempted, its old CPU being taken offline, resuming
+ * on some other CPU, then determining that its old CPU is now offline.
+ * It is OK to use RCU on an offline processor during initial boot, hence
+ * the check for rcu_scheduler_fully_active. Note also that it is OK
+ * for a CPU coming online to use RCU for one jiffy prior to marking itself
+ * online in the cpu_online_mask. Similarly, it is OK for a CPU going
+ * offline to continue to use RCU for one jiffy after marking itself
+ * offline in the cpu_online_mask. This leniency is necessary given the
+ * non-atomic nature of the online and offline processing, for example,
+ * the fact that a CPU enters the scheduler after completing the CPU_DYING
+ * notifiers.
+ *
+ * This is also why RCU internally marks CPUs online during the
+ * CPU_UP_PREPARE phase and offline during the CPU_DEAD phase.
+ *
+ * Disable checking if in an NMI handler because we cannot safely report
+ * errors from NMI handlers anyway.
+ */
+bool rcu_lockdep_current_cpu_online(void)
+{
+ struct rcu_data *rdp;
+ struct rcu_node *rnp;
+ bool ret;
+
+ if (in_nmi())
+ return 1;
+ preempt_disable();
+ rdp = &__get_cpu_var(rcu_sched_data);
+ rnp = rdp->mynode;
+ ret = (rdp->grpmask & rnp->qsmaskinit) ||
+ !rcu_scheduler_fully_active;
+ preempt_enable();
+ return ret;
+}
+EXPORT_SYMBOL_GPL(rcu_lockdep_current_cpu_online);
+
+#endif /* #ifdef CONFIG_HOTPLUG_CPU */
+
#endif /* #ifdef CONFIG_PROVE_RCU */
/**
return rcu_implicit_offline_qs(rdp);
}
+static int jiffies_till_stall_check(void)
+{
+ int till_stall_check = ACCESS_ONCE(rcu_cpu_stall_timeout);
+
+ /*
+ * Limit check must be consistent with the Kconfig limits
+ * for CONFIG_RCU_CPU_STALL_TIMEOUT.
+ */
+ if (till_stall_check < 3) {
+ ACCESS_ONCE(rcu_cpu_stall_timeout) = 3;
+ till_stall_check = 3;
+ } else if (till_stall_check > 300) {
+ ACCESS_ONCE(rcu_cpu_stall_timeout) = 300;
+ till_stall_check = 300;
+ }
+ return till_stall_check * HZ + RCU_STALL_DELAY_DELTA;
+}
+
static void record_gp_stall_check_time(struct rcu_state *rsp)
{
rsp->gp_start = jiffies;
- rsp->jiffies_stall = jiffies + RCU_SECONDS_TILL_STALL_CHECK;
+ rsp->jiffies_stall = jiffies + jiffies_till_stall_check();
}
static void print_other_cpu_stall(struct rcu_state *rsp)
raw_spin_unlock_irqrestore(&rnp->lock, flags);
return;
}
- rsp->jiffies_stall = jiffies + RCU_SECONDS_TILL_STALL_RECHECK;
-
- /*
- * Now rat on any tasks that got kicked up to the root rcu_node
- * due to CPU offlining.
- */
- ndetected = rcu_print_task_stall(rnp);
+ rsp->jiffies_stall = jiffies + 3 * jiffies_till_stall_check() + 3;
raw_spin_unlock_irqrestore(&rnp->lock, flags);
/*
* See Documentation/RCU/stallwarn.txt for info on how to debug
* RCU CPU stall warnings.
*/
- printk(KERN_ERR "INFO: %s detected stalls on CPUs/tasks: {",
+ printk(KERN_ERR "INFO: %s detected stalls on CPUs/tasks:",
rsp->name);
+ print_cpu_stall_info_begin();
rcu_for_each_leaf_node(rsp, rnp) {
raw_spin_lock_irqsave(&rnp->lock, flags);
ndetected += rcu_print_task_stall(rnp);
continue;
for (cpu = 0; cpu <= rnp->grphi - rnp->grplo; cpu++)
if (rnp->qsmask & (1UL << cpu)) {
- printk(" %d", rnp->grplo + cpu);
+ print_cpu_stall_info(rsp, rnp->grplo + cpu);
ndetected++;
}
}
- printk("} (detected by %d, t=%ld jiffies)\n",
+
+ /*
+ * Now rat on any tasks that got kicked up to the root rcu_node
+ * due to CPU offlining.
+ */
+ rnp = rcu_get_root(rsp);
+ raw_spin_lock_irqsave(&rnp->lock, flags);
+ ndetected = rcu_print_task_stall(rnp);
+ raw_spin_unlock_irqrestore(&rnp->lock, flags);
+
+ print_cpu_stall_info_end();
+ printk(KERN_CONT "(detected by %d, t=%ld jiffies)\n",
smp_processor_id(), (long)(jiffies - rsp->gp_start));
if (ndetected == 0)
printk(KERN_ERR "INFO: Stall ended before state dump start\n");
* See Documentation/RCU/stallwarn.txt for info on how to debug
* RCU CPU stall warnings.
*/
- printk(KERN_ERR "INFO: %s detected stall on CPU %d (t=%lu jiffies)\n",
- rsp->name, smp_processor_id(), jiffies - rsp->gp_start);
+ printk(KERN_ERR "INFO: %s self-detected stall on CPU", rsp->name);
+ print_cpu_stall_info_begin();
+ print_cpu_stall_info(rsp, smp_processor_id());
+ print_cpu_stall_info_end();
+ printk(KERN_CONT " (t=%lu jiffies)\n", jiffies - rsp->gp_start);
if (!trigger_all_cpu_backtrace())
dump_stack();
raw_spin_lock_irqsave(&rnp->lock, flags);
if (ULONG_CMP_GE(jiffies, rsp->jiffies_stall))
- rsp->jiffies_stall =
- jiffies + RCU_SECONDS_TILL_STALL_RECHECK;
+ rsp->jiffies_stall = jiffies +
+ 3 * jiffies_till_stall_check() + 3;
raw_spin_unlock_irqrestore(&rnp->lock, flags);
set_need_resched(); /* kick ourselves to get things going. */
rdp->passed_quiesce = 0;
} else
rdp->qs_pending = 0;
+ zero_cpu_stall_ticks(rdp);
}
}
*/
static void rcu_cleanup_dying_cpu(struct rcu_state *rsp)
{
- unsigned long flags;
int i;
unsigned long mask;
- int need_report;
int receive_cpu = cpumask_any(cpu_online_mask);
struct rcu_data *rdp = this_cpu_ptr(rsp->rda);
struct rcu_data *receive_rdp = per_cpu_ptr(rsp->rda, receive_cpu);
- struct rcu_node *rnp = rdp->mynode; /* For dying CPU. */
+ RCU_TRACE(struct rcu_node *rnp = rdp->mynode); /* For dying CPU. */
/* First, adjust the counts. */
if (rdp->nxtlist != NULL) {
"cpuofl");
rcu_report_qs_rdp(smp_processor_id(), rsp, rdp, rsp->gpnum);
/* Note that rcu_report_qs_rdp() might call trace_rcu_grace_period(). */
-
- /*
- * Remove the dying CPU from the bitmasks in the rcu_node
- * hierarchy. Because we are in stop_machine() context, we
- * automatically exclude ->onofflock critical sections.
- */
- do {
- raw_spin_lock_irqsave(&rnp->lock, flags);
- rnp->qsmaskinit &= ~mask;
- if (rnp->qsmaskinit != 0) {
- raw_spin_unlock_irqrestore(&rnp->lock, flags);
- break;
- }
- if (rnp == rdp->mynode) {
- need_report = rcu_preempt_offline_tasks(rsp, rnp, rdp);
- if (need_report & RCU_OFL_TASKS_NORM_GP)
- rcu_report_unblock_qs_rnp(rnp, flags);
- else
- raw_spin_unlock_irqrestore(&rnp->lock, flags);
- if (need_report & RCU_OFL_TASKS_EXP_GP)
- rcu_report_exp_rnp(rsp, rnp, true);
- } else
- raw_spin_unlock_irqrestore(&rnp->lock, flags);
- mask = rnp->grpmask;
- rnp = rnp->parent;
- } while (rnp != NULL);
}
/*
*/
static void rcu_cleanup_dead_cpu(int cpu, struct rcu_state *rsp)
{
+ unsigned long flags;
+ unsigned long mask;
+ int need_report = 0;
struct rcu_data *rdp = per_cpu_ptr(rsp->rda, cpu);
- struct rcu_node *rnp = rdp->mynode;
+ struct rcu_node *rnp = rdp->mynode; /* Outgoing CPU's rnp. */
+ /* Adjust any no-longer-needed kthreads. */
rcu_stop_cpu_kthread(cpu);
rcu_node_kthread_setaffinity(rnp, -1);
+
+ /* Remove the dying CPU from the bitmasks in the rcu_node hierarchy. */
+
+ /* Exclude any attempts to start a new grace period. */
+ raw_spin_lock_irqsave(&rsp->onofflock, flags);
+
+ /* Remove the outgoing CPU from the masks in the rcu_node hierarchy. */
+ mask = rdp->grpmask; /* rnp->grplo is constant. */
+ do {
+ raw_spin_lock(&rnp->lock); /* irqs already disabled. */
+ rnp->qsmaskinit &= ~mask;
+ if (rnp->qsmaskinit != 0) {
+ if (rnp != rdp->mynode)
+ raw_spin_unlock(&rnp->lock); /* irqs remain disabled. */
+ break;
+ }
+ if (rnp == rdp->mynode)
+ need_report = rcu_preempt_offline_tasks(rsp, rnp, rdp);
+ else
+ raw_spin_unlock(&rnp->lock); /* irqs remain disabled. */
+ mask = rnp->grpmask;
+ rnp = rnp->parent;
+ } while (rnp != NULL);
+
+ /*
+ * We still hold the leaf rcu_node structure lock here, and
+ * irqs are still disabled. The reason for this subterfuge is
+ * because invoking rcu_report_unblock_qs_rnp() with ->onofflock
+ * held leads to deadlock.
+ */
+ raw_spin_unlock(&rsp->onofflock); /* irqs remain disabled. */
+ rnp = rdp->mynode;
+ if (need_report & RCU_OFL_TASKS_NORM_GP)
+ rcu_report_unblock_qs_rnp(rnp, flags);
+ else
+ raw_spin_unlock_irqrestore(&rnp->lock, flags);
+ if (need_report & RCU_OFL_TASKS_EXP_GP)
+ rcu_report_exp_rnp(rsp, rnp, true);
}
#else /* #ifdef CONFIG_HOTPLUG_CPU */
void rcu_check_callbacks(int cpu, int user)
{
trace_rcu_utilization("Start scheduler-tick");
+ increment_cpu_stall_ticks();
if (user || rcu_is_cpu_rrupt_from_idle()) {
/*
}
EXPORT_SYMBOL_GPL(synchronize_rcu_bh);
+static atomic_t sync_sched_expedited_started = ATOMIC_INIT(0);
+static atomic_t sync_sched_expedited_done = ATOMIC_INIT(0);
+
+static int synchronize_sched_expedited_cpu_stop(void *data)
+{
+ /*
+ * There must be a full memory barrier on each affected CPU
+ * between the time that try_stop_cpus() is called and the
+ * time that it returns.
+ *
+ * In the current initial implementation of cpu_stop, the
+ * above condition is already met when the control reaches
+ * this point and the following smp_mb() is not strictly
+ * necessary. Do smp_mb() anyway for documentation and
+ * robustness against future implementation changes.
+ */
+ smp_mb(); /* See above comment block. */
+ return 0;
+}
+
+/*
+ * Wait for an rcu-sched grace period to elapse, but use "big hammer"
+ * approach to force grace period to end quickly. This consumes
+ * significant time on all CPUs, and is thus not recommended for
+ * any sort of common-case code.
+ *
+ * Note that it is illegal to call this function while holding any
+ * lock that is acquired by a CPU-hotplug notifier. Failing to
+ * observe this restriction will result in deadlock.
+ *
+ * This implementation can be thought of as an application of ticket
+ * locking to RCU, with sync_sched_expedited_started and
+ * sync_sched_expedited_done taking on the roles of the halves
+ * of the ticket-lock word. Each task atomically increments
+ * sync_sched_expedited_started upon entry, snapshotting the old value,
+ * then attempts to stop all the CPUs. If this succeeds, then each
+ * CPU will have executed a context switch, resulting in an RCU-sched
+ * grace period. We are then done, so we use atomic_cmpxchg() to
+ * update sync_sched_expedited_done to match our snapshot -- but
+ * only if someone else has not already advanced past our snapshot.
+ *
+ * On the other hand, if try_stop_cpus() fails, we check the value
+ * of sync_sched_expedited_done. If it has advanced past our
+ * initial snapshot, then someone else must have forced a grace period
+ * some time after we took our snapshot. In this case, our work is
+ * done for us, and we can simply return. Otherwise, we try again,
+ * but keep our initial snapshot for purposes of checking for someone
+ * doing our work for us.
+ *
+ * If we fail too many times in a row, we fall back to synchronize_sched().
+ */
+void synchronize_sched_expedited(void)
+{
+ int firstsnap, s, snap, trycount = 0;
+
+ /* Note that atomic_inc_return() implies full memory barrier. */
+ firstsnap = snap = atomic_inc_return(&sync_sched_expedited_started);
+ get_online_cpus();
+ WARN_ON_ONCE(cpu_is_offline(smp_processor_id()));
+
+ /*
+ * Each pass through the following loop attempts to force a
+ * context switch on each CPU.
+ */
+ while (try_stop_cpus(cpu_online_mask,
+ synchronize_sched_expedited_cpu_stop,
+ NULL) == -EAGAIN) {
+ put_online_cpus();
+
+ /* No joy, try again later. Or just synchronize_sched(). */
+ if (trycount++ < 10)
+ udelay(trycount * num_online_cpus());
+ else {
+ synchronize_sched();
+ return;
+ }
+
+ /* Check to see if someone else did our work for us. */
+ s = atomic_read(&sync_sched_expedited_done);
+ if (UINT_CMP_GE((unsigned)s, (unsigned)firstsnap)) {
+ smp_mb(); /* ensure test happens before caller kfree */
+ return;
+ }
+
+ /*
+ * Refetching sync_sched_expedited_started allows later
+ * callers to piggyback on our grace period. We subtract
+ * 1 to get the same token that the last incrementer got.
+ * We retry after they started, so our grace period works
+ * for them, and they started after our first try, so their
+ * grace period works for us.
+ */
+ get_online_cpus();
+ snap = atomic_read(&sync_sched_expedited_started);
+ smp_mb(); /* ensure read is before try_stop_cpus(). */
+ }
+
+ /*
+ * Everyone up to our most recent fetch is covered by our grace
+ * period. Update the counter, but only if our work is still
+ * relevant -- which it won't be if someone who started later
+ * than we did beat us to the punch.
+ */
+ do {
+ s = atomic_read(&sync_sched_expedited_done);
+ if (UINT_CMP_GE((unsigned)s, (unsigned)snap)) {
+ smp_mb(); /* ensure test happens before caller kfree */
+ break;
+ }
+ } while (atomic_cmpxchg(&sync_sched_expedited_done, s, snap) != s);
+
+ put_online_cpus();
+}
+EXPORT_SYMBOL_GPL(synchronize_sched_expedited);
+
/*
* Check to see if there is any immediate RCU-related work to be done
* by the current CPU, for the specified type of RCU, returning 1 if so.