We need full dynticks CPU to also be RCU nocb so
that we don't have to keep the tick to handle RCU
callbacks.

Make sure the range passed to nohz_full= boot
parameter is a subset of rcu_nocbs=

The CPUs that fail to meet this requirement will be
excluded from the nohz_full range. This is checked
early in boot time, before any CPU has the opportunity
to stop its tick.
Suggested-by: NSteven Rostedt <rostedt@goodmis.org>
Reviewed-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: NFrederic Weisbecker <fweisbec@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Chris Metcalf <cmetcalf@tilera.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Geoff Levand <geoff@infradead.org>
Cc: Gilad Ben Yossef <gilad@benyossef.com>
Cc: Hakan Akkan <hakanakkan@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Kevin Hilman <khilman@linaro.org>
Cc: Li Zhong <zhong@linux.vnet.ibm.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Thomas Gleixner <tglx@linutronix.de>

Adaptive-ticks CPUs inform RCU when they enter kernel mode, but they do
not necessarily turn the scheduler-clock tick back on.  This state of
affairs could result in RCU waiting on an adaptive-ticks CPU running
for an extended period in kernel mode.  Such a CPU will never run the
RCU state machine, and could therefore indefinitely extend the RCU state
machine, sooner or later resulting in an OOM condition.

This patch, inspired by an earlier patch by Frederic Weisbecker, therefore
causes RCU's force-quiescent-state processing to check for this condition
and to send an IPI to CPUs that remain in that state for too long.
"Too long" currently means about three jiffies by default, which is
quite some time for a CPU to remain in the kernel without blocking.
The rcu_tree.jiffies_till_first_fqs and rcutree.jiffies_till_next_fqs
sysfs variables may be used to tune "too long" if needed.
Reported-by: NFrederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>
Signed-off-by: NFrederic Weisbecker <fweisbec@gmail.com>
Cc: Chris Metcalf <cmetcalf@tilera.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Geoff Levand <geoff@infradead.org>
Cc: Gilad Ben Yossef <gilad@benyossef.com>
Cc: Hakan Akkan <hakanakkan@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Kevin Hilman <khilman@linaro.org>
Cc: Li Zhong <zhong@linux.vnet.ibm.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Thomas Gleixner <tglx@linutronix.de>

The as-documented rcu_nocb_poll will fail to enable this feature
for two reasons.  (1) there is an extra "s" in the documented
name which is not in the code, and (2) since it uses module_param,
it really is expecting a prefix, akin to "rcutree.fanout_leaf"
and the prefix isn't documented.

However, there are several reasons why we might not want to
simply fix the typo and add the prefix:

1) we'd end up with rcutree.rcu_nocb_poll, and rather probably make
a change to rcutree.nocb_poll

2) if we did #1, then the prefix wouldn't be consistent with the
rcu_nocbs=<cpumap> parameter (i.e. one with, one without prefix)

3) the use of module_param in a header file is less than desired,
since it isn't immediately obvious that it will get processed
via rcutree.c and get the prefix from that (although use of
module_param_named() could clarify that.)

4) the implied export of /sys/module/rcutree/parameters/rcu_nocb_poll
data to userspace via module_param() doesn't really buy us anything,
as it is read-only and we can tell if it is enabled already without
it, since there is a printk at early boot telling us so.

In light of all that, just change it from a module_param() to an
early_setup() call, and worry about adding it to /sys later on if
we decide to allow a dynamic setting of it.

Also change the variable to be tagged as read_mostly, since it
will only ever be fiddled with at most, once at boot.
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

The wait_event() at the head of the rcu_nocb_kthread() can result in
soft-lockup complaints if the CPU in question does not register RCU
callbacks for an extended period.  This commit therefore changes
the wait_event() to a wait_event_interruptible().
Reported-by: NFrederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

Currently, callback invocations from callback-free CPUs are accounted to
the CPU that registered the callback, but using the same field that is
used for normal callbacks.  This makes it impossible to determine from
debugfs output whether callbacks are in fact being diverted.  This commit
therefore adds a separate ->n_nocbs_invoked field in the rcu_data structure
in which diverted callback invocations are counted.  RCU's debugfs tracing
still displays normal callback invocations using ci=, but displayed
diverted callbacks with nci=.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

RCU callback execution can add significant OS jitter and also can
degrade both scheduling latency and, in asymmetric multiprocessors,
energy efficiency.  This commit therefore adds the ability for selected
CPUs ("rcu_nocbs=" boot parameter) to have their callbacks offloaded
to kthreads.  If the "rcu_nocb_poll" boot parameter is also specified,
these kthreads will do polling, removing the need for the offloaded
CPUs to do wakeups.  At least one CPU must be doing normal callback
processing: currently CPU 0 cannot be selected as a no-CBs CPU.
In addition, attempts to offline the last normal-CBs CPU will fail.

This feature was inspired by Jim Houston's and Joe Korty's JRCU, and
this commit includes fixes to problems located by Fengguang Wu's
kbuild test robot.

[ paulmck: Added gfp.h include file as suggested by Fengguang Wu. ]
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

This commit explicitly states the memory-ordering properties of the
RCU grace-period primitives.  Although these properties were in some
sense implied by the fundmental property of RCU ("a grace period must
wait for all pre-existing RCU read-side critical sections to complete"),
stating it explicitly will be a great labor-saving device.
Reported-by: NOleg Nesterov <oleg@redhat.com>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NOleg Nesterov <oleg@redhat.com>

The ->onofflock field in the rcu_state structure at one time synchronized
CPU-hotplug operations for RCU.  However, its scope has decreased over time
so that it now only protects the lists of orphaned RCU callbacks.  This
commit therefore renames it to ->orphan_lock to reflect its current use.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

There have been some embedded applications that would benefit from
use of expedited grace-period primitives.  In some ways, this is
similar to synchronize_net() doing either a normal or an expedited
grace period depending on lock state, but with control outside of
the kernel.

This commit therefore adds rcu_expedited boot and sysfs parameters
that cause the kernel to substitute expedited primitives for the
normal grace-period primitives.

[ paulmck: Add trace/event/rcu.h to kernel/srcu.c to avoid build error.
	   Get rid of infinite loop through contention path.]
Signed-off-by: NAntti P Miettinen <amiettinen@nvidia.com>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

The current implementation of RCU_FAST_NO_HZ tries reasonably hard to rid
the current CPU of RCU callbacks.  This is appropriate when the CPU is
entering idle, where it doesn't have much useful to do anyway, but is most
definitely not what you want when transitioning to user-mode execution.
This commit therefore detects the adaptive-tick case, and refrains from
burning CPU time getting rid of RCU callbacks in that case.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

The print_cpu_stall_fast_no_hz() function attempts to print -1 when
the ->idle_gp_timer is not pending, but unsigned arithmetic causes it
to instead print ULONG_MAX, which is 4294967295 on 32-bit systems and
18446744073709551615 on 64-bit systems.  Neither of these are the most
reader-friendly values, so this commit instead causes "timer not pending"
to be printed when ->idle_gp_timer is not pending.
Reported-by: NPaul Walmsley <paul@pwsan.com>
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

The rcu_print_detail_task_stall_rnp() function invokes
rcu_preempt_blocked_readers_cgp() to verify that there are some preempted
RCU readers blocking the current grace period outside of the protection
of the rcu_node structure's ->lock.  This means that the last blocked
reader might exit its RCU read-side critical section and remove itself
from the ->blkd_tasks list before the ->lock is acquired, resulting in
a segmentation fault when the subsequent code attempts to dereference
the now-NULL gp_tasks pointer.

This commit therefore moves the test under the lock.  This will not
have measurable effect on lock contention because this code is invoked
only when printing RCU CPU stall warnings, in other words, in the common
case, never.
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

The increment_cpu_stall_ticks() function listed each RCU flavor
explicitly, with an ifdef to handle preemptible RCU.  This commit
therefore applies for_each_rcu_flavor() to save a line of code.

Because this commit switches from a code-based enumeration of the
flavors of RCU to an rcu_state-list-based enumeration, it is no longer
possible to apply __get_cpu_var() to the per-CPU rcu_data structures.
We instead use __this_cpu_var() on the rcu_state structure's ->rda field
that references the corresponding rcu_data structures.
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

Commit 1217ed1b (rcu: permit rcu_read_unlock() to be called while holding
runqueue locks) made rcu_initiate_boost() restore irq state when releasing
the rcu_node structure's ->lock, but failed to update the header comment
accordingly.  This commit therefore brings the header comment up to date.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

The rcu_preempt_offline_tasks() moves all tasks queued on a given leaf
rcu_node structure to the root rcu_node, which is done when the last CPU
corresponding the the leaf rcu_node structure goes offline.  Now that
RCU-preempt's synchronize_rcu_expedited() implementation blocks CPU-hotplug
operations during the initialization of each rcu_node structure's
->boost_tasks pointer, rcu_preempt_offline_tasks() can do a better job
of setting the root rcu_node's ->boost_tasks pointer.

The key point is that rcu_preempt_offline_tasks() runs as part of the
CPU-hotplug process, so that a concurrent synchronize_rcu_expedited()
is guaranteed to either have not started on the one hand (in which case
there is no boosting on behalf of the expedited grace period) or to be
completely initialized on the other (in which case, in the absence of
other priority boosting, all ->boost_tasks pointers will be initialized).
Therefore, if rcu_preempt_offline_tasks() finds that the ->boost_tasks
pointer is equal to the ->exp_tasks pointer, it can be sure that it is
correctly placed.

In the case where there was boosting ongoing at the time that the
synchronize_rcu_expedited() function started, different nodes might start
boosting the tasks blocking the expedited grace period at different times.
In this mixed case, the root node will either be boosting tasks for
the expedited grace period already, or it will start as soon as it gets
done boosting for the normal grace period -- but in this latter case,
the root node's tasks needed to be boosted in any case.

This commit therefore adds a check of the ->boost_tasks pointer against
the ->exp_tasks pointer to the list that prevents updating ->boost_tasks.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

When rcu_preempt_offline_tasks() clears tasks from a leaf rcu_node
structure, it does not NULL out the structure's ->boost_tasks field.
This commit therefore fixes this issue.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

The current quiescent-state detection algorithm is needlessly
complex.  It records the grace-period number corresponding to
the quiescent state at the time of the quiescent state, which
works, but it seems better to simply erase any record of previous
quiescent states at the time that the CPU notices the new grace
period.  This has the further advantage of removing another piece
of RCU for which lockless reasoning is required.

Therefore, this commit makes this change.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

The synchronize_rcu_expedited() function disables interrupts across a
scan of all leaf rcu_node structures, which is not good for real-time
scheduling latency on large systems (hundreds or especially thousands
of CPUs).  This commit therefore holds off CPU-hotplug operations using
get_online_cpus(), and removes the prior acquisiion of the ->onofflock
(which required disabling interrupts).
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

In the C language, signed overflow is undefined.  It is true that
twos-complement arithmetic normally comes to the rescue, but if the
compiler can subvert this any time it has any information about the values
being compared.  For example, given "if (a - b > 0)", if the compiler
has enough information to realize that (for example) the value of "a"
is positive and that of "b" is negative, the compiler is within its
rights to optimize to a simple "if (1)", which might not be what you want.

This commit therefore converts synchronize_rcu_expedited()'s work-done
detection counter from signed to unsigned.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

As the first step towards allowing quiescent-state forcing to be
preemptible, this commit moves RCU quiescent-state forcing into the
same kthread that is now used to initialize and clean up after grace
periods.  This is yet another step towards keeping scheduling
latency down to a dull roar.

Updated to change from raw_spin_lock_irqsave() to raw_spin_lock_irq()
and to remove the now-unused rcu_state structure fields as suggested by
Peter Zijlstra.
Reported-by: NMike Galbraith <mgalbraith@suse.de>
Reported-by: NDimitri Sivanich <sivanich@sgi.com>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

In kernels built with CONFIG_RCU_FAST_NO_HZ=y, CPUs can accumulate a
large number of lazy callbacks, which as the name implies will be slow
to be invoked.  This can be a problem on small-memory systems, where the
default 6-second sleep for CPUs having only lazy RCU callbacks could well
be fatal.  This commit therefore installs an OOM hander that ensures that
every CPU with lazy callbacks has at least one non-lazy callback, in turn
ensuring timely advancement for these callbacks.

Updated to fix bug that disabled OOM killing, noted by Lai Jiangshan.

Updated to push the for_each_rcu_flavor() loop into rcu_oom_notify_cpu(),
thus reducing the number of IPIs, as suggested by Steven Rostedt.  Also
to make the for_each_online_cpu() loop be preemptible.  (Later, it might
be good to use smp_call_function(), as suggested by Peter Zijlstra.)
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: NSasha Levin <levinsasha928@gmail.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

Bring RCU into the new-age CPU-hotplug fold by modifying RCU's per-CPU
kthread code to use the new smp_hotplug_thread facility.

[ tglx: Adapted it to use callbacks and to the simplified rcu yield ]
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/r/20120716103948.673354828@linutronix.deSigned-off-by: NThomas Gleixner <tglx@linutronix.de>

The rcu_yield() code is amazing. It's there to avoid starvation of the
system when lots of (boosting) work is to be done.

Now looking at the code it's functionality is:

 Make the thread SCHED_OTHER and very nice, i.e. get it out of the way
 Arm a timer with 2 ticks
 schedule()

Now if the system goes idle the rcu task returns, regains SCHED_FIFO
and plugs on. If the systems stays busy the timer fires and wakes a
per node kthread which in turn makes the per cpu thread SCHED_FIFO and
brings it back on the cpu. For the boosting thread the "make it FIFO"
bit is missing and it just runs some magic boost checks. Now this is a
lot of code with extra threads and complexity.

It's way simpler to let the tasks when they detect overload schedule
away for 2 ticks and defer the normal wakeup as long as they are in
yielded state and the cpu is not idle.

That solves the same problem and the only difference is that when the
cpu goes idle it's not guaranteed that the thread returns right away,
but it won't be longer out than two ticks, so no harm is done. If
that's an issue than it is way simpler just to wake the task from
idle as RCU has callbacks there anyway.
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Namhyung Kim <namhyung@kernel.org>
Reviewed-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Link: http://lkml.kernel.org/r/20120716103948.131256723@linutronix.deSigned-off-by: NThomas Gleixner <tglx@linutronix.de>

The Linux kernel coding style says that single-statement blocks should
omit curly braces unless the other leg of the "if" statement has
multiple statements, in which case the curly braces should be included.
This commit fixes RCU's violations of this rule.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

If the nohz= boot parameter disables nohz, then RCU_FAST_NO_HZ needs to
also disable itself.  This commit therefore checks for tick_nohz_enabled
being zero, disabling rcu_prepare_for_idle() if so.  This commit assumes
that tick_nohz_enabled can change at runtime: If this is not the case,
then a simpler approach suffices.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

Currently, if several CPUs in the same package have all lazy RCU
callbacks, their wakeups will be uncorrelated.  If all the CPUs are in the
same power domain (as is often the case), this will result in unnecessary
power-ups of the package.  This commit therefore uses round_jiffies()
to round the timeouts to a second boundary, increasing the odds that
they can be coalesced with each other or with other timeouts.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

An uninitialized string may be displayed at the end of the rcu_preempt
detected stall info such as

0: (1 GPs behind) idle=075/140000000000000/0 =8?^D=8?^D
                                             ^^^^^^^^^^
if CONFIG_RCU_FAST_NO_HZ is not defined.

This trivial patch clears the string in this case.
Signed-off-by: NCarsten Emde <C.Emde@osadl.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

The CONFIG_TREE_PREEMPT_RCU and CONFIG_TINY_PREEMPT_RCU versions of
__rcu_read_lock() and __rcu_read_unlock() are identical, so this commit
consolidates them into kernel/rcupdate.h.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

The arrival of TREE_PREEMPT_RCU some years back included some ugly
code involving either #ifdef or #ifdef'ed wrapper functions to iterate
over all non-SRCU flavors of RCU.  This commit therefore introduces
a for_each_rcu_flavor() iterator over the rcu_state structures for each
flavor of RCU to clean up a bit of the ugliness.
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

With the advent of __this_cpu_ptr(), it is no longer necessary to pass
both the rcu_state and rcu_data structures into __rcu_process_callbacks().
This commit therefore computes the rcu_data pointer from the rcu_state
pointer within __rcu_process_callbacks() so that callers can pass in
only the pointer to the rcu_state structure.  This paves the way for
linking the rcu_state structures together and iterating over them.
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

This is a preparatory commit for increasing rcu_barrier()'s concurrency.
It adds a pointer in the rcu_data structure to the corresponding call_rcu()
function.  This allows a pointer to the rcu_data structure to imply the
function pointer, which allows _rcu_barrier() state to be placed in the
rcu_state structure.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: NJosh Triplett <josh@joshtriplett.org>

The rcu_node tree array is sized based on compile-time constants,
including NR_CPUS.  Although this approach has worked well in the past,
the recent trend by many distros to define NR_CPUS=4096 results in
excessive grace-period-initialization latencies.

This commit therefore substitutes the run-time computed nr_cpu_ids for
the compile-time NR_CPUS when building the tree.  This can result in
much of the compile-time-allocated rcu_node array being unused.  If
this is a major problem, you are in a specialized situation anyway,
so you can manually adjust the NR_CPUS, RCU_FANOUT, and RCU_FANOUT_LEAF
kernel config parameters.
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

Time to make the four-level-hierarchy setting less scary, so this
commit removes "Experimental" from the boot-time message.  Leave the
message in order to get a heads-up on any possible need to expand to
a five-level hierarchy.
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

Although making RCU_FANOUT_LEAF a kernel configuration parameter rather
than a fixed constant makes it easier for people to decrease cache-miss
overhead for large systems, it is of little help for people who must
run a single pre-built kernel binary.

This commit therefore allows the value of RCU_FANOUT_LEAF to be
increased (but not decreased!) via a boot-time parameter named
rcutree.rcu_fanout_leaf.
Reported-by: NMike Galbraith <efault@gmx.de>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

This reverts commit 616c310e.
(Move PREEMPT_RCU preemption to switch_to() invocation).
Testing by Sasha Levin <levinsasha928@gmail.com> showed that this
can result in deadlock due to invoking the scheduler when one of
the runqueue locks is held.  Because this commit was simply a
performance optimization, revert it.
Reported-by: NSasha Levin <levinsasha928@gmail.com>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: NSasha Levin <levinsasha928@gmail.com>

When a CPU is entering dyntick-idle mode, tick_nohz_stop_sched_tick()
calls rcu_needs_cpu() see if RCU needs that CPU, and, if not, computes the
next wakeup time based on the timer wheels.  Only later, when actually
entering the idle loop, rcu_prepare_for_idle() will be invoked.  In some
cases, rcu_prepare_for_idle() will post timers to wake the CPU back up.
But all for naught: The next wakeup time for the CPU has already been
computed, and posting a timer afterwards does not force that wakeup
time to be recomputed.  This means that rcu_prepare_for_idle()'s have
no effect.

This is not a problem on a busy system because something else will wake
up the CPU soon enough.  However, on lightly loaded systems, the CPU
might stay asleep for a considerable length of time.  If that CPU has
a callback that the rest of the system is waiting on, the system might
run very slowly or (in theory) even hang.

This commit avoids this problem by having rcu_needs_cpu() give
tick_nohz_stop_sched_tick() an estimate of when RCU will need the CPU
to wake back up, which tick_nohz_stop_sched_tick() takes into account
when programming the CPU's wakeup time.  An alternative approach is
for rcu_prepare_for_idle() to use hrtimers instead of normal timers,
but timers are much more efficient than are hrtimers for frequently
and repeatedly posting and cancelling a given timer, which is exactly
what RCU_FAST_NO_HZ does.
Reported-by: NPascal Chapperon <pascal.chapperon@wanadoo.fr>
Reported-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Tested-by: NPascal Chapperon <pascal.chapperon@wanadoo.fr>

The RCU_FAST_NO_HZ code relies on a number of per-CPU variables.
This works, but is hidden from someone scanning the data structures
in rcutree.h.  This commit therefore converts these per-CPU variables
to fields in the per-CPU rcu_dynticks structures.
Suggested-by: NPeter Zijlstra <peterz@infradead.org>
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Tested-by: NPascal Chapperon <pascal.chapperon@wanadoo.fr>

In the current code, a short dyntick-idle interval (where there is
at least one non-lazy callback on the CPU) and a long dyntick-idle
interval (where there are only lazy callbacks on the CPU) are traced
identically, which can be less than helpful.  This commit therefore
emits different event traces in these two cases.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Tested-by: NPascal Chapperon <pascal.chapperon@wanadoo.fr>

The current initialization of the RCU_FAST_NO_HZ per-CPU variables makes
needless and fragile assumptions about the initial value of things like
the jiffies counter.  This commit therefore explicitly initializes all of
them that are better started with a non-zero value.  It also adds some
comments describing the per-CPU state variables.
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>

The current RCU_FAST_NO_HZ assumes that timers do not migrate unless a
CPU goes offline, in which case it assumes that the CPU will have to come
out of dyntick-idle mode (cancelling the timer) in order to go offline.
This is important because when RCU_FAST_NO_HZ permits a CPU to enter
dyntick-idle mode despite having RCU callbacks pending, it posts a timer
on that CPU to force a wakeup on that CPU.  This wakeup ensures that the
CPU will eventually handle the end of the grace period, including invoking
its RCU callbacks.

However, Pascal Chapperon's test setup shows that the timer handler
rcu_idle_gp_timer_func() really does get invoked in some cases.  This is
problematic because this can cause the CPU that entered dyntick-idle
mode despite still having RCU callbacks pending to remain in
dyntick-idle mode indefinitely, which means that its RCU callbacks might
never be invoked.  This situation can result in grace-period delays or
even system hangs, which matches Pascal's observations of slow boot-up
and shutdown (https://lkml.org/lkml/2012/4/5/142).  See also the bugzilla:

	https://bugzilla.redhat.com/show_bug.cgi?id=806548

This commit therefore causes the "should never be invoked" timer handler
rcu_idle_gp_timer_func() to use smp_call_function_single() to wake up
the CPU for which the timer was intended, allowing that CPU to invoke
its RCU callbacks in a timely manner.
Reported-by: NPascal Chapperon <pascal.chapperon@wanadoo.fr>
Signed-off-by: NPaul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>