10 files changed, 446 insertions, 135 deletions

diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c
index e0e07fd55508..d566aba7e801 100644
--- a/kernel/context_tracking.c
+++ b/kernel/context_tracking.c
@@ -1,3 +1,19 @@
+/*
+ * Context tracking: Probe on high level context boundaries such as kernel
+ * and userspace. This includes syscalls and exceptions entry/exit.
+ *
+ * This is used by RCU to remove its dependency on the timer tick while a CPU
+ * runs in userspace.
+ *
+ *  Started by Frederic Weisbecker:
+ *
+ * Copyright (C) 2012 Red Hat, Inc., Frederic Weisbecker <fweisbec@redhat.com>
+ *
+ * Many thanks to Gilad Ben-Yossef, Paul McKenney, Ingo Molnar, Andrew Morton,
+ * Steven Rostedt, Peter Zijlstra for suggestions and improvements.
+ *
+ */
+
 #include <linux/context_tracking.h>
 #include <linux/rcupdate.h>
 #include <linux/sched.h>
@@ -6,8 +22,8 @@
 
 struct context_tracking {
         /*
-         * When active is false, hooks are not set to
-         * minimize overhead: TIF flags are cleared
+         * When active is false, probes are unset in order
+         * to minimize overhead: TIF flags are cleared
          * and calls to user_enter/exit are ignored. This
          * may be further optimized using static keys.
          */
@@ -24,6 +40,15 @@ static DEFINE_PER_CPU(struct context_tracking, context_tracking) = {
 #endif
 };
 
+/**
+ * user_enter - Inform the context tracking that the CPU is going to
+ *              enter userspace mode.
+ *
+ * This function must be called right before we switch from the kernel
+ * to userspace, when it's guaranteed the remaining kernel instructions
+ * to execute won't use any RCU read side critical section because this
+ * function sets RCU in extended quiescent state.
+ */
 void user_enter(void)
 {
         unsigned long flags;
@@ -39,40 +64,70 @@ void user_enter(void)
         if (in_interrupt())
                 return;
 
+        /* Kernel threads aren't supposed to go to userspace */
         WARN_ON_ONCE(!current->mm);
 
         local_irq_save(flags);
         if (__this_cpu_read(context_tracking.active) &&
             __this_cpu_read(context_tracking.state) != IN_USER) {
                 __this_cpu_write(context_tracking.state, IN_USER);
+                /*
+                 * At this stage, only low level arch entry code remains and
+                 * then we'll run in userspace. We can assume there won't be
+                 * any RCU read-side critical section until the next call to
+                 * user_exit() or rcu_irq_enter(). Let's remove RCU's dependency
+                 * on the tick.
+                 */
                 rcu_user_enter();
         }
         local_irq_restore(flags);
 }
 
+
+/**
+ * user_exit - Inform the context tracking that the CPU is
+ *             exiting userspace mode and entering the kernel.
+ *
+ * This function must be called after we entered the kernel from userspace
+ * before any use of RCU read side critical section. This potentially include
+ * any high level kernel code like syscalls, exceptions, signal handling, etc...
+ *
+ * This call supports re-entrancy. This way it can be called from any exception
+ * handler without needing to know if we came from userspace or not.
+ */
 void user_exit(void)
 {
         unsigned long flags;
 
-        /*
-         * Some contexts may involve an exception occuring in an irq,
-         * leading to that nesting:
-         * rcu_irq_enter() rcu_user_exit() rcu_user_exit() rcu_irq_exit()
-         * This would mess up the dyntick_nesting count though. And rcu_irq_*()
-         * helpers are enough to protect RCU uses inside the exception. So
-         * just return immediately if we detect we are in an IRQ.
-         */
         if (in_interrupt())
                 return;
 
         local_irq_save(flags);
         if (__this_cpu_read(context_tracking.state) == IN_USER) {
                 __this_cpu_write(context_tracking.state, IN_KERNEL);
+                /*
+                 * We are going to run code that may use RCU. Inform
+                 * RCU core about that (ie: we may need the tick again).
+                 */
                 rcu_user_exit();
         }
         local_irq_restore(flags);
 }
 
+
+/**
+ * context_tracking_task_switch - context switch the syscall callbacks
+ * @prev: the task that is being switched out
+ * @next: the task that is being switched in
+ *
+ * The context tracking uses the syscall slow path to implement its user-kernel
+ * boundaries probes on syscalls. This way it doesn't impact the syscall fast
+ * path on CPUs that don't do context tracking.
+ *
+ * But we need to clear the flag on the previous task because it may later
+ * migrate to some CPU that doesn't do the context tracking. As such the TIF
+ * flag may not be desired there.
+ */
 void context_tracking_task_switch(struct task_struct *prev,
                              struct task_struct *next)
 {
diff --git a/kernel/rcu.h b/kernel/rcu.h
index 20dfba576c2b..7f8e7590e3e5 100644
--- a/kernel/rcu.h
+++ b/kernel/rcu.h
@@ -111,4 +111,11 @@ static inline bool __rcu_reclaim(char *rn, struct rcu_head *head)
 
 extern int rcu_expedited;
 
+#ifdef CONFIG_RCU_STALL_COMMON
+
+extern int rcu_cpu_stall_suppress;
+int rcu_jiffies_till_stall_check(void);
+
+#endif /* #ifdef CONFIG_RCU_STALL_COMMON */
+
 #endif /* __LINUX_RCU_H */
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c
index a2cf76177b44..48ab70384a4c 100644
--- a/kernel/rcupdate.c
+++ b/kernel/rcupdate.c
@@ -404,11 +404,65 @@ EXPORT_SYMBOL_GPL(rcuhead_debug_descr);
 #endif /* #ifdef CONFIG_DEBUG_OBJECTS_RCU_HEAD */
 
 #if defined(CONFIG_TREE_RCU) || defined(CONFIG_TREE_PREEMPT_RCU) || defined(CONFIG_RCU_TRACE)
-void do_trace_rcu_torture_read(char *rcutorturename, struct rcu_head *rhp)
+void do_trace_rcu_torture_read(char *rcutorturename, struct rcu_head *rhp,
+                               unsigned long secs,
+                               unsigned long c_old, unsigned long c)
 {
-        trace_rcu_torture_read(rcutorturename, rhp);
+        trace_rcu_torture_read(rcutorturename, rhp, secs, c_old, c);
 }
 EXPORT_SYMBOL_GPL(do_trace_rcu_torture_read);
 #else
-#define do_trace_rcu_torture_read(rcutorturename, rhp) do { } while (0)
+#define do_trace_rcu_torture_read(rcutorturename, rhp, secs, c_old, c) \
+        do { } while (0)
 #endif
+
+#ifdef CONFIG_RCU_STALL_COMMON
+
+#ifdef CONFIG_PROVE_RCU
+#define RCU_STALL_DELAY_DELTA          (5 * HZ)
+#else
+#define RCU_STALL_DELAY_DELTA          0
+#endif
+
+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);
+
+int rcu_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 int rcu_panic(struct notifier_block *this, unsigned long ev, void *ptr)
+{
+        rcu_cpu_stall_suppress = 1;
+        return NOTIFY_DONE;
+}
+
+static struct notifier_block rcu_panic_block = {
+        .notifier_call = rcu_panic,
+};
+
+static int __init check_cpu_stall_init(void)
+{
+        atomic_notifier_chain_register(&panic_notifier_list, &rcu_panic_block);
+        return 0;
+}
+early_initcall(check_cpu_stall_init);
+
+#endif /* #ifdef CONFIG_RCU_STALL_COMMON */
diff --git a/kernel/rcutiny.c b/kernel/rcutiny.c
index e7dce58f9c2a..a0714a51b6d7 100644
--- a/kernel/rcutiny.c
+++ b/kernel/rcutiny.c
@@ -51,10 +51,10 @@ static void __call_rcu(struct rcu_head *head,
                        void (*func)(struct rcu_head *rcu),
                        struct rcu_ctrlblk *rcp);
 
-#include "rcutiny_plugin.h"
-
 static long long rcu_dynticks_nesting = DYNTICK_TASK_EXIT_IDLE;
 
+#include "rcutiny_plugin.h"
+
 /* Common code for rcu_idle_enter() and rcu_irq_exit(), see kernel/rcutree.c. */
 static void rcu_idle_enter_common(long long newval)
 {
@@ -193,7 +193,7 @@ EXPORT_SYMBOL(rcu_is_cpu_idle);
  * interrupts don't count, we must be running at the first interrupt
  * level.
  */
-int rcu_is_cpu_rrupt_from_idle(void)
+static int rcu_is_cpu_rrupt_from_idle(void)
 {
         return rcu_dynticks_nesting <= 1;
 }
@@ -205,6 +205,7 @@ int rcu_is_cpu_rrupt_from_idle(void)
  */
 static int rcu_qsctr_help(struct rcu_ctrlblk *rcp)
 {
+        reset_cpu_stall_ticks(rcp);
         if (rcp->rcucblist != NULL &&
             rcp->donetail != rcp->curtail) {
                 rcp->donetail = rcp->curtail;
@@ -251,6 +252,7 @@ void rcu_bh_qs(int cpu)
  */
 void rcu_check_callbacks(int cpu, int user)
 {
+        check_cpu_stalls();
         if (user || rcu_is_cpu_rrupt_from_idle())
                 rcu_sched_qs(cpu);
         else if (!in_softirq())
diff --git a/kernel/rcutiny_plugin.h b/kernel/rcutiny_plugin.h
index f85016a2309b..8a233002faeb 100644
--- a/kernel/rcutiny_plugin.h
+++ b/kernel/rcutiny_plugin.h
@@ -33,6 +33,9 @@ struct rcu_ctrlblk {
         struct rcu_head **donetail;     /* ->next pointer of last "done" CB. */
         struct rcu_head **curtail;      /* ->next pointer of last CB. */
         RCU_TRACE(long qlen);           /* Number of pending CBs. */
+        RCU_TRACE(unsigned long gp_start); /* Start time for stalls. */
+        RCU_TRACE(unsigned long ticks_this_gp); /* Statistic for stalls. */
+        RCU_TRACE(unsigned long jiffies_stall); /* Jiffies at next stall. */
         RCU_TRACE(char *name);          /* Name of RCU type. */
 };
 
@@ -54,6 +57,51 @@ int rcu_scheduler_active __read_mostly;
 EXPORT_SYMBOL_GPL(rcu_scheduler_active);
 #endif /* #ifdef CONFIG_DEBUG_LOCK_ALLOC */
 
+#ifdef CONFIG_RCU_TRACE
+
+static void check_cpu_stall(struct rcu_ctrlblk *rcp)
+{
+        unsigned long j;
+        unsigned long js;
+
+        if (rcu_cpu_stall_suppress)
+                return;
+        rcp->ticks_this_gp++;
+        j = jiffies;
+        js = rcp->jiffies_stall;
+        if (*rcp->curtail && ULONG_CMP_GE(j, js)) {
+                pr_err("INFO: %s stall on CPU (%lu ticks this GP) idle=%llx (t=%lu jiffies q=%ld)\n",
+                       rcp->name, rcp->ticks_this_gp, rcu_dynticks_nesting,
+                       jiffies - rcp->gp_start, rcp->qlen);
+                dump_stack();
+        }
+        if (*rcp->curtail && ULONG_CMP_GE(j, js))
+                rcp->jiffies_stall = jiffies +
+                        3 * rcu_jiffies_till_stall_check() + 3;
+        else if (ULONG_CMP_GE(j, js))
+                rcp->jiffies_stall = jiffies + rcu_jiffies_till_stall_check();
+}
+
+static void check_cpu_stall_preempt(void);
+
+#endif /* #ifdef CONFIG_RCU_TRACE */
+
+static void reset_cpu_stall_ticks(struct rcu_ctrlblk *rcp)
+{
+#ifdef CONFIG_RCU_TRACE
+        rcp->ticks_this_gp = 0;
+        rcp->gp_start = jiffies;
+        rcp->jiffies_stall = jiffies + rcu_jiffies_till_stall_check();
+#endif /* #ifdef CONFIG_RCU_TRACE */
+}
+
+static void check_cpu_stalls(void)
+{
+        RCU_TRACE(check_cpu_stall(&rcu_bh_ctrlblk));
+        RCU_TRACE(check_cpu_stall(&rcu_sched_ctrlblk));
+        RCU_TRACE(check_cpu_stall_preempt());
+}
+
 #ifdef CONFIG_TINY_PREEMPT_RCU
 
 #include <linux/delay.h>
@@ -448,6 +496,7 @@ static void rcu_preempt_start_gp(void)
                 /* Official start of GP. */
                 rcu_preempt_ctrlblk.gpnum++;
                 RCU_TRACE(rcu_preempt_ctrlblk.n_grace_periods++);
+                reset_cpu_stall_ticks(&rcu_preempt_ctrlblk.rcb);
 
                 /* Any blocked RCU readers block new GP. */
                 if (rcu_preempt_blocked_readers_any())
@@ -1054,4 +1103,11 @@ MODULE_AUTHOR("Paul E. McKenney");
 MODULE_DESCRIPTION("Read-Copy Update tracing for tiny implementation");
 MODULE_LICENSE("GPL");
 
+static void check_cpu_stall_preempt(void)
+{
+#ifdef CONFIG_TINY_PREEMPT_RCU
+        check_cpu_stall(&rcu_preempt_ctrlblk.rcb);
+#endif /* #ifdef CONFIG_TINY_PREEMPT_RCU */
+}
+
 #endif /* #ifdef CONFIG_RCU_TRACE */
diff --git a/kernel/rcutorture.c b/kernel/rcutorture.c
index 31dea01c85fd..e1f3a8c96724 100644
--- a/kernel/rcutorture.c
+++ b/kernel/rcutorture.c
@@ -46,6 +46,7 @@
 #include <linux/stat.h>
 #include <linux/srcu.h>
 #include <linux/slab.h>
+#include <linux/trace_clock.h>
 #include <asm/byteorder.h>
 
 MODULE_LICENSE("GPL");
@@ -207,6 +208,20 @@ MODULE_PARM_DESC(rcutorture_runnable, "Start rcutorture at boot");
 #define rcu_can_boost() 0
 #endif /* #else #if defined(CONFIG_RCU_BOOST) && !defined(CONFIG_HOTPLUG_CPU) */
 
+#ifdef CONFIG_RCU_TRACE
+static u64 notrace rcu_trace_clock_local(void)
+{
+        u64 ts = trace_clock_local();
+        unsigned long __maybe_unused ts_rem = do_div(ts, NSEC_PER_USEC);
+        return ts;
+}
+#else /* #ifdef CONFIG_RCU_TRACE */
+static u64 notrace rcu_trace_clock_local(void)
+{
+        return 0ULL;
+}
+#endif /* #else #ifdef CONFIG_RCU_TRACE */
+
 static unsigned long shutdown_time;     /* jiffies to system shutdown. */
 static unsigned long boost_starttime;   /* jiffies of next boost test start. */
 DEFINE_MUTEX(boost_mutex);              /* protect setting boost_starttime */
@@ -845,7 +860,7 @@ static int rcu_torture_boost(void *arg)
                 /* Wait for the next test interval. */
                 oldstarttime = boost_starttime;
                 while (ULONG_CMP_LT(jiffies, oldstarttime)) {
-                        schedule_timeout_uninterruptible(1);
+                        schedule_timeout_interruptible(oldstarttime - jiffies);
                         rcu_stutter_wait("rcu_torture_boost");
                         if (kthread_should_stop() ||
                             fullstop != FULLSTOP_DONTSTOP)
@@ -1028,7 +1043,6 @@ void rcutorture_trace_dump(void)
                 return;
         if (atomic_xchg(&beenhere, 1) != 0)
                 return;
-        do_trace_rcu_torture_read(cur_ops->name, (struct rcu_head *)~0UL);
         ftrace_dump(DUMP_ALL);
 }
 
@@ -1042,13 +1056,16 @@ static void rcu_torture_timer(unsigned long unused)
 {
         int idx;
         int completed;
+        int completed_end;
         static DEFINE_RCU_RANDOM(rand);
         static DEFINE_SPINLOCK(rand_lock);
         struct rcu_torture *p;
         int pipe_count;
+        unsigned long long ts;
 
         idx = cur_ops->readlock();
         completed = cur_ops->completed();
+        ts = rcu_trace_clock_local();
         p = rcu_dereference_check(rcu_torture_current,
                                   rcu_read_lock_bh_held() ||
                                   rcu_read_lock_sched_held() ||
@@ -1058,7 +1075,6 @@ static void rcu_torture_timer(unsigned long unused)
                 cur_ops->readunlock(idx);
                 return;
         }
-        do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu);
         if (p->rtort_mbtest == 0)
                 atomic_inc(&n_rcu_torture_mberror);
         spin_lock(&rand_lock);
@@ -1071,10 +1087,14 @@ static void rcu_torture_timer(unsigned long unused)
                 /* Should not happen, but... */
                 pipe_count = RCU_TORTURE_PIPE_LEN;
         }
-        if (pipe_count > 1)
+        completed_end = cur_ops->completed();
+        if (pipe_count > 1) {
+                do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu, ts,
+                                          completed, completed_end);
                 rcutorture_trace_dump();
+        }
         __this_cpu_inc(rcu_torture_count[pipe_count]);
-        completed = cur_ops->completed() - completed;
+        completed = completed_end - completed;
         if (completed > RCU_TORTURE_PIPE_LEN) {
                 /* Should not happen, but... */
                 completed = RCU_TORTURE_PIPE_LEN;
@@ -1094,11 +1114,13 @@ static int
 rcu_torture_reader(void *arg)
 {
         int completed;
+        int completed_end;
         int idx;
         DEFINE_RCU_RANDOM(rand);
         struct rcu_torture *p;
         int pipe_count;
         struct timer_list t;
+        unsigned long long ts;
 
         VERBOSE_PRINTK_STRING("rcu_torture_reader task started");
         set_user_nice(current, 19);
@@ -1112,6 +1134,7 @@ rcu_torture_reader(void *arg)
                 }
                 idx = cur_ops->readlock();
                 completed = cur_ops->completed();
+                ts = rcu_trace_clock_local();
                 p = rcu_dereference_check(rcu_torture_current,
                                           rcu_read_lock_bh_held() ||
                                           rcu_read_lock_sched_held() ||
@@ -1122,7 +1145,6 @@ rcu_torture_reader(void *arg)
                         schedule_timeout_interruptible(HZ);
                         continue;
                 }
-                do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu);
                 if (p->rtort_mbtest == 0)
                         atomic_inc(&n_rcu_torture_mberror);
                 cur_ops->read_delay(&rand);
@@ -1132,10 +1154,14 @@ rcu_torture_reader(void *arg)
                         /* Should not happen, but... */
                         pipe_count = RCU_TORTURE_PIPE_LEN;
                 }
-                if (pipe_count > 1)
+                completed_end = cur_ops->completed();
+                if (pipe_count > 1) {
+                        do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu,
+                                                  ts, completed, completed_end);
                         rcutorture_trace_dump();
+                }
                 __this_cpu_inc(rcu_torture_count[pipe_count]);
-                completed = cur_ops->completed() - completed;
+                completed = completed_end - completed;
                 if (completed > RCU_TORTURE_PIPE_LEN) {
                         /* Should not happen, but... */
                         completed = RCU_TORTURE_PIPE_LEN;
@@ -1301,19 +1327,35 @@ static void rcu_torture_shuffle_tasks(void)
                                 set_cpus_allowed_ptr(reader_tasks[i],
                                                      shuffle_tmp_mask);
         }
-
         if (fakewriter_tasks) {
                 for (i = 0; i < nfakewriters; i++)
                         if (fakewriter_tasks[i])
                                 set_cpus_allowed_ptr(fakewriter_tasks[i],
                                                      shuffle_tmp_mask);
         }
-
         if (writer_task)
                 set_cpus_allowed_ptr(writer_task, shuffle_tmp_mask);
-
         if (stats_task)
                 set_cpus_allowed_ptr(stats_task, shuffle_tmp_mask);
+        if (stutter_task)
+                set_cpus_allowed_ptr(stutter_task, shuffle_tmp_mask);
+        if (fqs_task)
+                set_cpus_allowed_ptr(fqs_task, shuffle_tmp_mask);
+        if (shutdown_task)
+                set_cpus_allowed_ptr(shutdown_task, shuffle_tmp_mask);
+#ifdef CONFIG_HOTPLUG_CPU
+        if (onoff_task)
+                set_cpus_allowed_ptr(onoff_task, shuffle_tmp_mask);
+#endif /* #ifdef CONFIG_HOTPLUG_CPU */
+        if (stall_task)
+                set_cpus_allowed_ptr(stall_task, shuffle_tmp_mask);
+        if (barrier_cbs_tasks)
+                for (i = 0; i < n_barrier_cbs; i++)
+                        if (barrier_cbs_tasks[i])
+                                set_cpus_allowed_ptr(barrier_cbs_tasks[i],
+                                                     shuffle_tmp_mask);
+        if (barrier_task)
+                set_cpus_allowed_ptr(barrier_task, shuffle_tmp_mask);
 
         if (rcu_idle_cpu == -1)
                 rcu_idle_cpu = num_online_cpus() - 1;
@@ -1749,7 +1791,7 @@ static int rcu_torture_barrier_init(void)
         barrier_cbs_wq =
                 kzalloc(n_barrier_cbs * sizeof(barrier_cbs_wq[0]),
                         GFP_KERNEL);
-        if (barrier_cbs_tasks == NULL || barrier_cbs_wq == 0)
+        if (barrier_cbs_tasks == NULL || !barrier_cbs_wq)
                 return -ENOMEM;
         for (i = 0; i < n_barrier_cbs; i++) {
                 init_waitqueue_head(&barrier_cbs_wq[i]);
diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index e441b77b614e..5b8ad827fd86 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -105,7 +105,7 @@ int rcu_num_nodes __read_mostly = NUM_RCU_NODES; /* Total # rcu_nodes in use. */
  * The rcu_scheduler_active variable transitions from zero to one just
  * before the first task is spawned.  So when this variable is zero, RCU
  * can assume that there is but one task, allowing RCU to (for example)
- * optimized synchronize_sched() to a simple barrier().  When this variable
+ * optimize synchronize_sched() to a simple barrier().  When this variable
  * is one, RCU must actually do all the hard work required to detect real
  * grace periods.  This variable is also used to suppress boot-time false
  * positives from lockdep-RCU error checking.
@@ -217,12 +217,6 @@ module_param(blimit, long, 0444);
 module_param(qhimark, long, 0444);
 module_param(qlowmark, long, 0444);
 
-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 ulong jiffies_till_first_fqs = RCU_JIFFIES_TILL_FORCE_QS;
 static ulong jiffies_till_next_fqs = RCU_JIFFIES_TILL_FORCE_QS;
 
@@ -305,17 +299,27 @@ cpu_has_callbacks_ready_to_invoke(struct rcu_data *rdp)
 }
 
 /*
- * Does the current CPU require a yet-as-unscheduled grace period?
+ * Does the current CPU require a not-yet-started grace period?
+ * The caller must have disabled interrupts to prevent races with
+ * normal callback registry.
  */
 static int
 cpu_needs_another_gp(struct rcu_state *rsp, struct rcu_data *rdp)
 {
-        struct rcu_head **ntp;
+        int i;
 
-        ntp = rdp->nxttail[RCU_DONE_TAIL +
-                           (ACCESS_ONCE(rsp->completed) != rdp->completed)];
-        return rdp->nxttail[RCU_DONE_TAIL] && ntp && *ntp &&
-               !rcu_gp_in_progress(rsp);
+        if (rcu_gp_in_progress(rsp))
+                return 0;  /* No, a grace period is already in progress. */
+        if (!rdp->nxttail[RCU_NEXT_TAIL])
+                return 0;  /* No, this is a no-CBs (or offline) CPU. */
+        if (*rdp->nxttail[RCU_NEXT_READY_TAIL])
+                return 1;  /* Yes, this CPU has newly registered callbacks. */
+        for (i = RCU_WAIT_TAIL; i < RCU_NEXT_TAIL; i++)
+                if (rdp->nxttail[i - 1] != rdp->nxttail[i] &&
+                    ULONG_CMP_LT(ACCESS_ONCE(rsp->completed),
+                                 rdp->nxtcompleted[i]))
+                        return 1;  /* Yes, CBs for future grace period. */
+        return 0; /* No grace period needed. */
 }
 
 /*
@@ -336,7 +340,7 @@ static struct rcu_node *rcu_get_root(struct rcu_state *rsp)
 static void rcu_eqs_enter_common(struct rcu_dynticks *rdtp, long long oldval,
                                 bool user)
 {
-        trace_rcu_dyntick("Start", oldval, 0);
+        trace_rcu_dyntick("Start", oldval, rdtp->dynticks_nesting);
         if (!user && !is_idle_task(current)) {
                 struct task_struct *idle = idle_task(smp_processor_id());
 
@@ -727,7 +731,7 @@ EXPORT_SYMBOL_GPL(rcu_lockdep_current_cpu_online);
  * interrupt from idle, return true.  The caller must have at least
  * disabled preemption.
  */
-int rcu_is_cpu_rrupt_from_idle(void)
+static int rcu_is_cpu_rrupt_from_idle(void)
 {
         return __get_cpu_var(rcu_dynticks).dynticks_nesting <= 1;
 }
@@ -793,28 +797,10 @@ static int rcu_implicit_dynticks_qs(struct rcu_data *rdp)
         return 0;
 }
 
-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 + jiffies_till_stall_check();
+        rsp->jiffies_stall = jiffies + rcu_jiffies_till_stall_check();
 }
 
 /*
@@ -857,7 +843,7 @@ static void print_other_cpu_stall(struct rcu_state *rsp)
                 raw_spin_unlock_irqrestore(&rnp->lock, flags);
                 return;
         }
-        rsp->jiffies_stall = jiffies + 3 * jiffies_till_stall_check() + 3;
+        rsp->jiffies_stall = jiffies + 3 * rcu_jiffies_till_stall_check() + 3;
         raw_spin_unlock_irqrestore(&rnp->lock, flags);
 
         /*
@@ -935,7 +921,7 @@ static void print_cpu_stall(struct rcu_state *rsp)
         raw_spin_lock_irqsave(&rnp->lock, flags);
         if (ULONG_CMP_GE(jiffies, rsp->jiffies_stall))
                 rsp->jiffies_stall = jiffies +
-                                     3 * jiffies_till_stall_check() + 3;
+                                     3 * rcu_jiffies_till_stall_check() + 3;
         raw_spin_unlock_irqrestore(&rnp->lock, flags);
 
         set_need_resched();  /* kick ourselves to get things going. */
@@ -966,12 +952,6 @@ static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp)
         }
 }
 
-static int rcu_panic(struct notifier_block *this, unsigned long ev, void *ptr)
-{
-        rcu_cpu_stall_suppress = 1;
-        return NOTIFY_DONE;
-}
-
 /**
  * rcu_cpu_stall_reset - prevent further stall warnings in current grace period
  *
@@ -989,15 +969,6 @@ void rcu_cpu_stall_reset(void)
                 rsp->jiffies_stall = jiffies + ULONG_MAX / 2;
 }
 
-static struct notifier_block rcu_panic_block = {
-        .notifier_call = rcu_panic,
-};
-
-static void __init check_cpu_stall_init(void)
-{
-        atomic_notifier_chain_register(&panic_notifier_list, &rcu_panic_block);
-}
-
 /*
  * Update CPU-local rcu_data state to record the newly noticed grace period.
  * This is used both when we started the grace period and when we notice
@@ -1071,6 +1042,145 @@ static void init_callback_list(struct rcu_data *rdp)
 }
 
 /*
+ * Determine the value that ->completed will have at the end of the
+ * next subsequent grace period.  This is used to tag callbacks so that
+ * a CPU can invoke callbacks in a timely fashion even if that CPU has
+ * been dyntick-idle for an extended period with callbacks under the
+ * influence of RCU_FAST_NO_HZ.
+ *
+ * The caller must hold rnp->lock with interrupts disabled.
+ */
+static unsigned long rcu_cbs_completed(struct rcu_state *rsp,
+                                       struct rcu_node *rnp)
+{
+        /*
+         * If RCU is idle, we just wait for the next grace period.
+         * But we can only be sure that RCU is idle if we are looking
+         * at the root rcu_node structure -- otherwise, a new grace
+         * period might have started, but just not yet gotten around
+         * to initializing the current non-root rcu_node structure.
+         */
+        if (rcu_get_root(rsp) == rnp && rnp->gpnum == rnp->completed)
+                return rnp->completed + 1;
+
+        /*
+         * Otherwise, wait for a possible partial grace period and
+         * then the subsequent full grace period.
+         */
+        return rnp->completed + 2;
+}
+
+/*
+ * If there is room, assign a ->completed number to any callbacks on
+ * this CPU that have not already been assigned.  Also accelerate any
+ * callbacks that were previously assigned a ->completed number that has
+ * since proven to be too conservative, which can happen if callbacks get
+ * assigned a ->completed number while RCU is idle, but with reference to
+ * a non-root rcu_node structure.  This function is idempotent, so it does
+ * not hurt to call it repeatedly.
+ *
+ * The caller must hold rnp->lock with interrupts disabled.
+ */
+static void rcu_accelerate_cbs(struct rcu_state *rsp, struct rcu_node *rnp,
+                               struct rcu_data *rdp)
+{
+        unsigned long c;
+        int i;
+
+        /* If the CPU has no callbacks, nothing to do. */
+        if (!rdp->nxttail[RCU_NEXT_TAIL] || !*rdp->nxttail[RCU_DONE_TAIL])
+                return;
+
+        /*
+         * Starting from the sublist containing the callbacks most
+         * recently assigned a ->completed number and working down, find the
+         * first sublist that is not assignable to an upcoming grace period.
+         * Such a sublist has something in it (first two tests) and has
+         * a ->completed number assigned that will complete sooner than
+         * the ->completed number for newly arrived callbacks (last test).
+         *
+         * The key point is that any later sublist can be assigned the
+         * same ->completed number as the newly arrived callbacks, which
+         * means that the callbacks in any of these later sublist can be
+         * grouped into a single sublist, whether or not they have already
+         * been assigned a ->completed number.
+         */
+        c = rcu_cbs_completed(rsp, rnp);
+        for (i = RCU_NEXT_TAIL - 1; i > RCU_DONE_TAIL; i--)
+                if (rdp->nxttail[i] != rdp->nxttail[i - 1] &&
+                    !ULONG_CMP_GE(rdp->nxtcompleted[i], c))
+                        break;
+
+        /*
+         * If there are no sublist for unassigned callbacks, leave.
+         * At the same time, advance "i" one sublist, so that "i" will
+         * index into the sublist where all the remaining callbacks should
+         * be grouped into.
+         */
+        if (++i >= RCU_NEXT_TAIL)
+                return;
+
+        /*
+         * Assign all subsequent callbacks' ->completed number to the next
+         * full grace period and group them all in the sublist initially
+         * indexed by "i".
+         */
+        for (; i <= RCU_NEXT_TAIL; i++) {
+                rdp->nxttail[i] = rdp->nxttail[RCU_NEXT_TAIL];
+                rdp->nxtcompleted[i] = c;
+        }
+
+        /* Trace depending on how much we were able to accelerate. */
+        if (!*rdp->nxttail[RCU_WAIT_TAIL])
+                trace_rcu_grace_period(rsp->name, rdp->gpnum, "AccWaitCB");
+        else
+                trace_rcu_grace_period(rsp->name, rdp->gpnum, "AccReadyCB");
+}
+
+/*
+ * Move any callbacks whose grace period has completed to the
+ * RCU_DONE_TAIL sublist, then compact the remaining sublists and
+ * assign ->completed numbers to any callbacks in the RCU_NEXT_TAIL
+ * sublist.  This function is idempotent, so it does not hurt to
+ * invoke it repeatedly.  As long as it is not invoked -too- often...
+ *
+ * The caller must hold rnp->lock with interrupts disabled.
+ */
+static void rcu_advance_cbs(struct rcu_state *rsp, struct rcu_node *rnp,
+                            struct rcu_data *rdp)
+{
+        int i, j;
+
+        /* If the CPU has no callbacks, nothing to do. */
+        if (!rdp->nxttail[RCU_NEXT_TAIL] || !*rdp->nxttail[RCU_DONE_TAIL])
+                return;
+
+        /*
+         * Find all callbacks whose ->completed numbers indicate that they
+         * are ready to invoke, and put them into the RCU_DONE_TAIL sublist.
+         */
+        for (i = RCU_WAIT_TAIL; i < RCU_NEXT_TAIL; i++) {
+                if (ULONG_CMP_LT(rnp->completed, rdp->nxtcompleted[i]))
+                        break;
+                rdp->nxttail[RCU_DONE_TAIL] = rdp->nxttail[i];
+        }
+        /* Clean up any sublist tail pointers that were misordered above. */
+        for (j = RCU_WAIT_TAIL; j < i; j++)
+                rdp->nxttail[j] = rdp->nxttail[RCU_DONE_TAIL];
+
+        /* Copy down callbacks to fill in empty sublists. */
+        for (j = RCU_WAIT_TAIL; i < RCU_NEXT_TAIL; i++, j++) {
+                if (rdp->nxttail[j] == rdp->nxttail[RCU_NEXT_TAIL])
+                        break;
+                rdp->nxttail[j] = rdp->nxttail[i];
+                rdp->nxtcompleted[j] = rdp->nxtcompleted[i];
+        }
+
+        /* Classify any remaining callbacks. */
+        rcu_accelerate_cbs(rsp, rnp, rdp);
+}
+
+/*
  * Advance this CPU's callbacks, but only if the current grace period
  * has ended.  This may be called only from the CPU to whom the rdp
  * belongs.  In addition, the corresponding leaf rcu_node structure's
@@ -1080,12 +1190,15 @@ static void
 __rcu_process_gp_end(struct rcu_state *rsp, struct rcu_node *rnp, struct rcu_data *rdp)
 {
         /* Did another grace period end? */
-        if (rdp->completed != rnp->completed) {
+        if (rdp->completed == rnp->completed) {
 
-                /* Advance callbacks.  No harm if list empty. */
-                rdp->nxttail[RCU_DONE_TAIL] = rdp->nxttail[RCU_WAIT_TAIL];
-                rdp->nxttail[RCU_WAIT_TAIL] = rdp->nxttail[RCU_NEXT_READY_TAIL];
-                rdp->nxttail[RCU_NEXT_READY_TAIL] = rdp->nxttail[RCU_NEXT_TAIL];
+                /* No, so just accelerate recent callbacks. */
+                rcu_accelerate_cbs(rsp, rnp, rdp);
+
+        } else {
+
+                /* Advance callbacks. */
+                rcu_advance_cbs(rsp, rnp, rdp);
 
                 /* Remember that we saw this grace-period completion. */
                 rdp->completed = rnp->completed;
@@ -1392,17 +1505,10 @@ rcu_start_gp(struct rcu_state *rsp, unsigned long flags)
         /*
          * Because there is no grace period in progress right now,
          * any callbacks we have up to this point will be satisfied
-         * by the next grace period.  So promote all callbacks to be
-         * handled after the end of the next grace period.  If the
-         * CPU is not yet aware of the end of the previous grace period,
-         * we need to allow for the callback advancement that will
-         * occur when it does become aware.  Deadlock prevents us from
-         * making it aware at this point: We cannot acquire a leaf
-         * rcu_node ->lock while holding the root rcu_node ->lock.
+         * by the next grace period.  So this is a good place to
+         * assign a grace period number to recently posted callbacks.
          */
-        rdp->nxttail[RCU_NEXT_READY_TAIL] = rdp->nxttail[RCU_NEXT_TAIL];
-        if (rdp->completed == rsp->completed)
-                rdp->nxttail[RCU_WAIT_TAIL] = rdp->nxttail[RCU_NEXT_TAIL];
+        rcu_accelerate_cbs(rsp, rnp, rdp);
 
         rsp->gp_flags = RCU_GP_FLAG_INIT;
         raw_spin_unlock(&rnp->lock); /* Interrupts remain disabled. */
@@ -1527,7 +1633,7 @@ rcu_report_qs_rdp(int cpu, struct rcu_state *rsp, struct rcu_data *rdp)
                  * This GP can't end until cpu checks in, so all of our
                  * callbacks can be processed during the next GP.
                  */
-                rdp->nxttail[RCU_NEXT_READY_TAIL] = rdp->nxttail[RCU_NEXT_TAIL];
+                rcu_accelerate_cbs(rsp, rnp, rdp);
 
                 rcu_report_qs_rnp(mask, rsp, rnp, flags); /* rlses rnp->lock */
         }
@@ -1779,7 +1885,7 @@ static void rcu_do_batch(struct rcu_state *rsp, struct rcu_data *rdp)
         long bl, count, count_lazy;
         int i;
 
-        /* If no callbacks are ready, just return.*/
+        /* If no callbacks are ready, just return. */
         if (!cpu_has_callbacks_ready_to_invoke(rdp)) {
                 trace_rcu_batch_start(rsp->name, rdp->qlen_lazy, rdp->qlen, 0);
                 trace_rcu_batch_end(rsp->name, 0, !!ACCESS_ONCE(rdp->nxtlist),
@@ -2008,19 +2114,19 @@ __rcu_process_callbacks(struct rcu_state *rsp)
 
         WARN_ON_ONCE(rdp->beenonline == 0);
 
-        /*
-         * Advance callbacks in response to end of earlier grace
-         * period that some other CPU ended.
-         */
+        /* Handle the end of a grace period that some other CPU ended.  */
         rcu_process_gp_end(rsp, rdp);
 
         /* Update RCU state based on any recent quiescent states. */
         rcu_check_quiescent_state(rsp, rdp);
 
         /* Does this CPU require a not-yet-started grace period? */
+        local_irq_save(flags);
         if (cpu_needs_another_gp(rsp, rdp)) {
-                raw_spin_lock_irqsave(&rcu_get_root(rsp)->lock, flags);
+                raw_spin_lock(&rcu_get_root(rsp)->lock); /* irqs disabled. */
                 rcu_start_gp(rsp, flags);  /* releases above lock */
+        } else {
+                local_irq_restore(flags);
         }
 
         /* If there are callbacks ready, invoke them. */
@@ -2719,9 +2825,6 @@ rcu_boot_init_percpu_data(int cpu, struct rcu_state *rsp)
         rdp->dynticks = &per_cpu(rcu_dynticks, cpu);
         WARN_ON_ONCE(rdp->dynticks->dynticks_nesting != DYNTICK_TASK_EXIT_IDLE);
         WARN_ON_ONCE(atomic_read(&rdp->dynticks->dynticks) != 1);
-#ifdef CONFIG_RCU_USER_QS
-        WARN_ON_ONCE(rdp->dynticks->in_user);
-#endif
         rdp->cpu = cpu;
         rdp->rsp = rsp;
         rcu_boot_init_nocb_percpu_data(rdp);
@@ -2938,6 +3041,10 @@ static void __init rcu_init_one(struct rcu_state *rsp,
 
         BUILD_BUG_ON(MAX_RCU_LVLS > ARRAY_SIZE(buf));  /* Fix buf[] init! */
 
+        /* Silence gcc 4.8 warning about array index out of range. */
+        if (rcu_num_lvls > RCU_NUM_LVLS)
+                panic("rcu_init_one: rcu_num_lvls overflow");
+
         /* Initialize the level-tracking arrays. */
 
         for (i = 0; i < rcu_num_lvls; i++)
@@ -3074,7 +3181,6 @@ void __init rcu_init(void)
         cpu_notifier(rcu_cpu_notify, 0);
         for_each_online_cpu(cpu)
                 rcu_cpu_notify(NULL, CPU_UP_PREPARE, (void *)(long)cpu);
-        check_cpu_stall_init();
 }
 
 #include "rcutree_plugin.h"
diff --git a/kernel/rcutree.h b/kernel/rcutree.h
index 4b69291b093d..c896b5045d9d 100644
--- a/kernel/rcutree.h
+++ b/kernel/rcutree.h
@@ -102,10 +102,6 @@ struct rcu_dynticks {
                                     /* idle-period nonlazy_posted snapshot. */
         int tick_nohz_enabled_snap; /* Previously seen value from sysfs. */
 #endif /* #ifdef CONFIG_RCU_FAST_NO_HZ */
-#ifdef CONFIG_RCU_USER_QS
-        bool ignore_user_qs;        /* Treat userspace as extended QS or not */
-        bool in_user;               /* Is the CPU in userland from RCU POV? */
-#endif
 };
 
 /* RCU's kthread states for tracing. */
@@ -282,6 +278,8 @@ struct rcu_data {
          */
         struct rcu_head *nxtlist;
         struct rcu_head **nxttail[RCU_NEXT_SIZE];
+        unsigned long   nxtcompleted[RCU_NEXT_SIZE];
+                                        /* grace periods for sublists. */
         long            qlen_lazy;      /* # of lazy queued callbacks */
         long            qlen;           /* # of queued callbacks, incl lazy */
         long            qlen_last_fqs_check;
@@ -343,11 +341,6 @@ struct rcu_data {
 
 #define RCU_JIFFIES_TILL_FORCE_QS        3      /* for rsp->jiffies_force_qs */
 
-#ifdef CONFIG_PROVE_RCU
-#define RCU_STALL_DELAY_DELTA          (5 * HZ)
-#else
-#define RCU_STALL_DELAY_DELTA          0
-#endif
 #define RCU_STALL_RAT_DELAY             2       /* Allow other CPUs time */
                                                 /*  to take at least one */
                                                 /*  scheduling clock irq */
diff --git a/kernel/srcu.c b/kernel/srcu.c
index 2b859828cdc3..01d5ccb8bfe3 100644
--- a/kernel/srcu.c
+++ b/kernel/srcu.c
@@ -282,12 +282,8 @@ static int srcu_readers_active(struct srcu_struct *sp)
  */
 void cleanup_srcu_struct(struct srcu_struct *sp)
 {
-        int sum;
-
-        sum = srcu_readers_active(sp);
-        WARN_ON(sum);  /* Leakage unless caller handles error. */
-        if (sum != 0)
-                return;
+        if (WARN_ON(srcu_readers_active(sp)))
+                return; /* Leakage unless caller handles error. */
         free_percpu(sp->per_cpu_ref);
         sp->per_cpu_ref = NULL;
 }
@@ -302,9 +298,8 @@ int __srcu_read_lock(struct srcu_struct *sp)
 {
         int idx;
 
+        idx = ACCESS_ONCE(sp->completed) & 0x1;
         preempt_disable();
-        idx = rcu_dereference_index_check(sp->completed,
-                                          rcu_read_lock_sched_held()) & 0x1;
         ACCESS_ONCE(this_cpu_ptr(sp->per_cpu_ref)->c[idx]) += 1;
         smp_mb(); /* B */  /* Avoid leaking the critical section. */
         ACCESS_ONCE(this_cpu_ptr(sp->per_cpu_ref)->seq[idx]) += 1;
@@ -321,10 +316,8 @@ EXPORT_SYMBOL_GPL(__srcu_read_lock);
  */
 void __srcu_read_unlock(struct srcu_struct *sp, int idx)
 {
-        preempt_disable();
         smp_mb(); /* C */  /* Avoid leaking the critical section. */
-        ACCESS_ONCE(this_cpu_ptr(sp->per_cpu_ref)->c[idx]) -= 1;
-        preempt_enable();
+        this_cpu_dec(sp->per_cpu_ref->c[idx]);
 }
 EXPORT_SYMBOL_GPL(__srcu_read_unlock);
 
@@ -423,6 +416,7 @@ static void __synchronize_srcu(struct srcu_struct *sp, int trycount)
                            !lock_is_held(&rcu_sched_lock_map),
                            "Illegal synchronize_srcu() in same-type SRCU (or RCU) read-side critical section");
 
+        might_sleep();
         init_completion(&rcu.completion);
 
         head->next = NULL;
@@ -455,10 +449,12 @@ static void __synchronize_srcu(struct srcu_struct *sp, int trycount)
  * synchronize_srcu - wait for prior SRCU read-side critical-section completion
  * @sp: srcu_struct with which to synchronize.
  *
- * Flip the completed counter, and wait for the old count to drain to zero.
- * As with classic RCU, the updater must use some separate means of
- * synchronizing concurrent updates.  Can block; must be called from
- * process context.
+ * Wait for the count to drain to zero of both indexes. To avoid the
+ * possible starvation of synchronize_srcu(), it waits for the count of
+ * the index=((->completed & 1) ^ 1) to drain to zero at first,
+ * and then flip the completed and wait for the count of the other index.
+ *
+ * Can block; must be called from process context.
  *
  * Note that it is illegal to call synchronize_srcu() from the corresponding
  * SRCU read-side critical section; doing so will result in deadlock.
@@ -480,12 +476,11 @@ EXPORT_SYMBOL_GPL(synchronize_srcu);
  * Wait for an SRCU grace period to elapse, but be more aggressive about
  * spinning rather than blocking when waiting.
  *
- * Note that it is illegal to call this function while holding any lock
- * that is acquired by a CPU-hotplug notifier.  It is also illegal to call
- * synchronize_srcu_expedited() from the corresponding SRCU read-side
- * critical section; doing so will result in deadlock.  However, it is
- * perfectly legal to call synchronize_srcu_expedited() on one srcu_struct
- * from some other srcu_struct's read-side critical section, as long as
+ * Note that it is also illegal to call synchronize_srcu_expedited()
+ * from the corresponding SRCU read-side critical section;
+ * doing so will result in deadlock.  However, it is perfectly legal
+ * to call synchronize_srcu_expedited() on one srcu_struct from some
+ * other srcu_struct's read-side critical section, as long as
  * the resulting graph of srcu_structs is acyclic.
  */
 void synchronize_srcu_expedited(struct srcu_struct *sp)
diff --git a/kernel/trace/trace_clock.c b/kernel/trace/trace_clock.c
index 394783531cbb..1bbb1b200cec 100644
--- a/kernel/trace/trace_clock.c
+++ b/kernel/trace/trace_clock.c
@@ -44,6 +44,7 @@ u64 notrace trace_clock_local(void)
 
         return clock;
 }
+EXPORT_SYMBOL_GPL(trace_clock_local);
 
 /*
  * trace_clock(): 'between' trace clock. Not completely serialized,