Because the Tasks RCU ->rtp_exit_list is initialized at rcu_init()
time while there is only one CPU running with interrupts disabled, it
is not possible for an exiting task to encounter an uninitialized list.
This commit therefore replaces the conditional initialization with
a WARN_ON_ONCE().
Reported-by: Frederic Weisbecker <frederic@kernel.org>
Closes: https://lore.kernel.org/all/ZdiNXmO3wRvmzPsr@lothringen/
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Tasks Trace RCU needs a single-byte cmpxchg(), but no such thing exists.
Therefore, rcu_trc_cmpxchg_need_qs() emulates one using field substitution
and a four-byte cmpxchg(), such that the other three bytes are always
atomically updated to their old values. This works, but results in
false-positive KCSAN failures because as far as KCSAN knows, this
cmpxchg() operation is updating all four bytes.
This commit therefore encloses the cmpxchg() in a data_race() and adds
a single-byte instrument_atomic_read_write(), thus telling KCSAN exactly
what is going on so as to avoid the false positives.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Marco Elver <elver@google.com>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Currently, there are rcu_data structure fields named ->rcu_onl_gp_seq
and ->rcu_ofl_gp_seq that track the rcu_state.gp_flags field at the
time of the corresponding CPU's last online or offline operation,
respectively. However, this information is not particularly useful.
It would be better to instead track the grace period state kept
in rcu_state.gp_state. This would also be consistent with the
initialization in rcu_boot_init_percpu_data(), which is to RCU_GP_CLEANED
(an rcu_state.gp_state value), and also with the diagnostics in
rcu_implicit_dynticks_qs(), whose format is consistent with an integer,
not a bitmask.
This commit therefore makes this change and changes the names to
->rcu_onl_gp_flags and ->rcu_ofl_gp_flags, respectively.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
The rcu_state.n_online_cpus value is only ever updated by CPU-hotplug
operations, which are serialized. However, this value is read locklessly.
This commit therefore marks those reads. While in the area, it also
adds ASSERT_EXCLUSIVE_WRITER() calls just in case parallel CPU hotplug
becomes a thing.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
The rcu_sync structure's ->gp_count field is updated under the protection
of ->rss_lock, but read locklessly, and KCSAN noted the data race.
This commit therefore uses WRITE_ONCE() to do this update to clearly
document its racy nature.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
This commit adds READ_ONCE() to a lockless diagnostic read from
rcu_state.gp_flags to avoid giving the compiler any chance whatsoever
of confusing the diagnostic state printed.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Because Tiny RCU is used only in kernels built with either
CONFIG_PREEMPT_NONE=y or CONFIG_PREEMPT_VOLUNTARY=y, there has not been
any need for TINY RCU to explicitly disable preemption. However, the
prospect of lazy preemption changes that, and preemption means that
the non-atomic increment in synchronize_rcu() can be preempted, with
the possibility that one of the increments is lost. This could cause
failures for users of the APIs that poll RCU grace periods.
This commit therefore adds the needed preempt_disable() and
preempt_enable() call to Tiny RCU.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Ankur Arora <ankur.a.arora@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
The TINY_RCU rcu_process_callbacks() function is only ever invoked from
a softirq handler, which means that BH is already disabled. This commit
therefore removes the redundant local_bh_disable() and local_bh_ennable()
from this function.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Currently, if a Kconfig option depends on TASKS_RCU, it conditionally does
"select TASKS_RCU if PREEMPTION". This works, but requires any change in
this enablement logic to be replicated across all such "select" clauses.
This commit therefore creates a new NEED_TASKS_RCU Kconfig option so
that the default value of TASKS_RCU can depend on a combination of this
new option and any needed enablement logic, so that this logic is in
one place.
While in the area, also anticipate a likely future change by adding
PREEMPT_AUTO to that logic.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Ankur Arora <ankur.a.arora@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Because Tiny SRCU is used only in kernels built with either
CONFIG_PREEMPT_NONE=y or CONFIG_PREEMPT_VOLUNTARY=y, there has not
been any need for TINY SRCU to explicitly disable preemption. However,
the prospect of lazy preemption changes that, and the lazy-preemption
patches do result in rcutorture runs finding both too-short grace periods
and grace-period hangs for Tiny SRCU.
This commit therefore adds the needed preempt_disable() and
preempt_enable() calls to Tiny SRCU.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Ankur Arora <ankur.a.arora@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Right now, TINY_RCU depends on (!PREEMPTION && !SMP), which has served the
kernel well for many years due to the fact that PREEMPT_RCU is normally
a synonym for PREEMPTION. But with the advent of lazy preemption,
it will be possible to have non-preemptible RCU in a preemptible kernel,
so that kernels could be built with PREEMPT_RCU=n and PREEMPTION=y.
This commit therefore makes TINY_RCU depend on (!PREEMPT_RCU && !SMP),
thus allowing for a non-preemptible RCU in preemptible kernels.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Ankur Arora <ankur.a.arora@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
The synchronize_rcu() call is going to be reworked, thus
this patch adds dedicated fields into the rcu_state structure.
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Currently, all waits for grace periods sleep at TASK_UNINTERRUPTIBLE,
regardless of RCU flavor. This has worked well, but there have been
cases where a longer-than-average Tasks RCU grace period has triggered
softlockup splats, many of them, before the Tasks RCU CPU stall warning
appears. These softlockup splats unnecessarily consume console bandwidth
and complicate diagnosis of the underlying problem. Plus a long but not
pathologically long Tasks RCU grace period might trigger a few softlockup
splats before completing normally, which generates noise for no good
reason.
This commit therefore causes Tasks RCU grace periods to sleep at TASK_IDLE
priority. If there really is a persistent problem, the eventual Tasks
RCU CPU stall warning will flag it, and without the extra noise.
Reported-by: Breno Leitao <leitao@debian.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
It turns out that only one CPU at a time will ever invoke
rcu_torture_pipe_update_one() on a given rcu_torture structure.
This commit therefore adds three ASSERT_EXCLUSIVE_WRITER() calls
to enlist KCSAN's aid in checking this.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
If a callback flood prevents grace period from completing, rcutorture
does a WARN_ON(). Avoiding this WARN_ON() currently requires that at
least three grace periods elapse during an eight-second callback-flood
interval. Unfortunately, the current debug information does not include
anything about the grace-period state. This commit therefore adds a
call to cur_ops->gp_kthread_dbg(), if this function pointer is non-NULL.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
This commit adds the number of online CPUs to the state dump following
an unsuccesful callback-flood test.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
There is some indications that rcu_softirq_qs() might be more generally
used than anticipated. This commit therefore adds some lockdep assertions
and some cautionary tales in a new kernel-doc header.
Link: https://lore.kernel.org/all/Zd4DXTyCf17lcTfq@debian.debian/
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Yan Zhai <yan@cloudflare.com>
Cc: <netdev@vger.kernel.org>
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
The current code will scan the entirety of each per-CPU list of exiting
tasks in ->rtp_exit_list with interrupts disabled. This is normally just
fine, because each CPU typically won't have very many tasks in this state.
However, if a large number of tasks block late in do_exit(), these lists
could be arbitrarily long. Low probability, perhaps, but it really
could happen.
This commit therefore occasionally re-enables interrupts while traversing
these lists, inserting a dummy element to hold the current place in the
list. In kernels built with CONFIG_PREEMPT_RT=y, this re-enabling happens
after each list element is processed, otherwise every one-to-two jiffies.
[ paulmck: Apply Frederic Weisbecker feedback. ]
Link: https://lore.kernel.org/all/ZdeI_-RfdLR8jlsm@localhost.localdomain/
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Sebastian Siewior <bigeasy@linutronix.de>
Cc: Anna-Maria Behnsen <anna-maria@linutronix.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Holding a mutex across synchronize_rcu_tasks() and acquiring
that same mutex in code called from do_exit() after its call to
exit_tasks_rcu_start() but before its call to exit_tasks_rcu_stop()
results in deadlock. This is by design, because tasks that are far
enough into do_exit() are no longer present on the tasks list, making
it a bit difficult for RCU Tasks to find them, let alone wait on them
to do a voluntary context switch. However, such deadlocks are becoming
more frequent. In addition, lockdep currently does not detect such
deadlocks and they can be difficult to reproduce.
In addition, if a task voluntarily context switches during that time
(for example, if it blocks acquiring a mutex), then this task is in an
RCU Tasks quiescent state. And with some adjustments, RCU Tasks could
just as well take advantage of that fact.
This commit therefore eliminates these deadlock by replacing the
SRCU-based wait for do_exit() completion with per-CPU lists of tasks
currently exiting. A given task will be on one of these per-CPU lists for
the same period of time that this task would previously have been in the
previous SRCU read-side critical section. These lists enable RCU Tasks
to find the tasks that have already been removed from the tasks list,
but that must nevertheless be waited upon.
The RCU Tasks grace period gathers any of these do_exit() tasks that it
must wait on, and adds them to the list of holdouts. Per-CPU locking
and get_task_struct() are used to synchronize addition to and removal
from these lists.
Link: https://lore.kernel.org/all/20240118021842.290665-1-chenzhongjin@huawei.com/
Reported-by: Chen Zhongjin <chenzhongjin@huawei.com>
Reported-by: Yang Jihong <yangjihong1@huawei.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Tested-by: Yang Jihong <yangjihong1@huawei.com>
Tested-by: Chen Zhongjin <chenzhongjin@huawei.com>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
This commit continues the elimination of deadlocks involving do_exit()
and RCU tasks by causing exit_tasks_rcu_start() to add the current
task to a per-CPU list and causing exit_tasks_rcu_stop() to remove the
current task from whatever list it is on. These lists will be used to
track tasks that are exiting, while still accounting for any RCU-tasks
quiescent states that these tasks pass though.
[ paulmck: Apply Frederic Weisbecker feedback. ]
Link: https://lore.kernel.org/all/20240118021842.290665-1-chenzhongjin@huawei.com/
Reported-by: Chen Zhongjin <chenzhongjin@huawei.com>
Reported-by: Yang Jihong <yangjihong1@huawei.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Tested-by: Yang Jihong <yangjihong1@huawei.com>
Tested-by: Chen Zhongjin <chenzhongjin@huawei.com>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Holding a mutex across synchronize_rcu_tasks() and acquiring
that same mutex in code called from do_exit() after its call to
exit_tasks_rcu_start() but before its call to exit_tasks_rcu_stop()
results in deadlock. This is by design, because tasks that are far
enough into do_exit() are no longer present on the tasks list, making
it a bit difficult for RCU Tasks to find them, let alone wait on them
to do a voluntary context switch. However, such deadlocks are becoming
more frequent. In addition, lockdep currently does not detect such
deadlocks and they can be difficult to reproduce.
In addition, if a task voluntarily context switches during that time
(for example, if it blocks acquiring a mutex), then this task is in an
RCU Tasks quiescent state. And with some adjustments, RCU Tasks could
just as well take advantage of that fact.
This commit therefore initializes the data structures that will be needed
to rely on these quiescent states and to eliminate these deadlocks.
Link: https://lore.kernel.org/all/20240118021842.290665-1-chenzhongjin@huawei.com/
Reported-by: Chen Zhongjin <chenzhongjin@huawei.com>
Reported-by: Yang Jihong <yangjihong1@huawei.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Tested-by: Yang Jihong <yangjihong1@huawei.com>
Tested-by: Chen Zhongjin <chenzhongjin@huawei.com>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
In order for RCU Tasks to reliably maintain per-CPU lists of exiting
tasks, those lists must be initialized before it is possible for tasks
to exit, especially given that the boot CPU is not necessarily CPU 0
(an example being, powerpc kexec() kernels). And at the time that
rcu_init_tasks_generic() is called, a task could potentially exit,
unconventional though that sort of thing might be.
This commit therefore moves the calls to cblist_init_generic() from
functions called from rcu_init_tasks_generic() to a new function named
tasks_cblist_init_generic() that is invoked from rcu_init().
This constituted a bug in a commit that never went to mainline, so
there is no need for any backporting to -stable.
Reported-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Holding a mutex across synchronize_rcu_tasks() and acquiring
that same mutex in code called from do_exit() after its call to
exit_tasks_rcu_start() but before its call to exit_tasks_rcu_stop()
results in deadlock. This is by design, because tasks that are far
enough into do_exit() are no longer present on the tasks list, making
it a bit difficult for RCU Tasks to find them, let alone wait on them
to do a voluntary context switch. However, such deadlocks are becoming
more frequent. In addition, lockdep currently does not detect such
deadlocks and they can be difficult to reproduce.
In addition, if a task voluntarily context switches during that time
(for example, if it blocks acquiring a mutex), then this task is in an
RCU Tasks quiescent state. And with some adjustments, RCU Tasks could
just as well take advantage of that fact.
This commit therefore adds the data structures that will be needed
to rely on these quiescent states and to eliminate these deadlocks.
Link: https://lore.kernel.org/all/20240118021842.290665-1-chenzhongjin@huawei.com/
Reported-by: Chen Zhongjin <chenzhongjin@huawei.com>
Reported-by: Yang Jihong <yangjihong1@huawei.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Tested-by: Yang Jihong <yangjihong1@huawei.com>
Tested-by: Chen Zhongjin <chenzhongjin@huawei.com>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
With commit '6a010a49b63a ("cgroup: Make !percpu threadgroup_rwsem
operations optional")' usage of rcu_sync_enter_start is removed.
So this function can also be removed.
In the words of Oleg Nesterov:
__rcu_sync_enter(wait => false) is a better alternative if
someone needs rcu_sync_enter_start() again.
Link: https://lore.kernel.org/all/20220725121208.GB28662@redhat.com/
Signed-off-by: Onkarnath <onkarnath.1@samsung.com>
Signed-off-by: Maninder Singh <maninder1.s@samsung.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Currently, if rcu_torture_writer() sees fewer than ten grace periods
having elapsed during a call to stutter_wait() that actually waited,
the rtort_pipe_count warning is emitted. This has worked well for
a long time. Except that the rcutorture TREE07 scenario now does a
short-term 14-second RCU CPU stall, which can most definitely case
false-positive rtort_pipe_count warnings.
This commit therefore changes rcu_torture_writer() to compute the
full expected holdoff and stall duration, and to refuse to report any
rtort_pipe_count warnings until after all stalls have completed.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Cc: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
The comments added in commit 1ef990c4b36b ("srcu: No need to
advance/accelerate if no callback enqueued") are a bit confusing.
The comments are describing a scenario for code that was moved and is
no longer the way it was (snapshot after advancing). Improve the code
comments to reflect this and also document why acceleration can never
fail.
Cc: Frederic Weisbecker <frederic@kernel.org>
Cc: Neeraj Upadhyay <neeraj.iitr10@gmail.com>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
To allow more flexible arrangements while still provide a single kernel
for distros, provide a boot time parameter to enable/disable lazy RCU.
Specify:
rcutree.enable_rcu_lazy=[y|1|n|0]
Which also requires
rcu_nocbs=all
at boot time to enable/disable lazy RCU.
To disable it by default at build time when CONFIG_RCU_LAZY=y, the new
CONFIG_RCU_LAZY_DEFAULT_OFF can be used.
Signed-off-by: Qais Yousef (Google) <qyousef@layalina.io>
Tested-by: Andrea Righi <andrea.righi@canonical.com>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
The variable name jiffies_till_flush is too generic and therefore:
* It may shadow a global variable
* It doesn't tell on what it operates
Make the name more precise, along with the related APIs.
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
TREE04 running on short iterations can produce writer stalls of the
following kind:
??? Writer stall state RTWS_EXP_SYNC(4) g3968 f0x0 ->state 0x2 cpu 0
task:rcu_torture_wri state:D stack:14568 pid:83 ppid:2 flags:0x00004000
Call Trace:
<TASK>
__schedule+0x2de/0x850
? trace_event_raw_event_rcu_exp_funnel_lock+0x6d/0xb0
schedule+0x4f/0x90
synchronize_rcu_expedited+0x430/0x670
? __pfx_autoremove_wake_function+0x10/0x10
? __pfx_synchronize_rcu_expedited+0x10/0x10
do_rtws_sync.constprop.0+0xde/0x230
rcu_torture_writer+0x4b4/0xcd0
? __pfx_rcu_torture_writer+0x10/0x10
kthread+0xc7/0xf0
? __pfx_kthread+0x10/0x10
ret_from_fork+0x2f/0x50
? __pfx_kthread+0x10/0x10
ret_from_fork_asm+0x1b/0x30
</TASK>
Waiting for an expedited grace period and polling for an expedited
grace period both are operations that internally rely on the same
workqueue performing necessary asynchronous work.
However, a dependency chain is involved between those two operations,
as depicted below:
====== CPU 0 ======= ====== CPU 1 =======
synchronize_rcu_expedited()
exp_funnel_lock()
mutex_lock(&rcu_state.exp_mutex);
start_poll_synchronize_rcu_expedited
queue_work(rcu_gp_wq, &rnp->exp_poll_wq);
synchronize_rcu_expedited_queue_work()
queue_work(rcu_gp_wq, &rew->rew_work);
wait_event() // A, wait for &rew->rew_work completion
mutex_unlock() // B
//======> switch to kworker
sync_rcu_do_polled_gp() {
synchronize_rcu_expedited()
exp_funnel_lock()
mutex_lock(&rcu_state.exp_mutex); // C, wait B
....
} // D
Since workqueues are usually implemented on top of several kworkers
handling the queue concurrently, the above situation wouldn't deadlock
most of the time because A then doesn't depend on D. But in case of
memory stress, a single kworker may end up handling alone all the works
in a serialized way. In that case the above layout becomes a problem
because A then waits for D, closing a circular dependency:
A -> D -> C -> B -> A
This however only happens when CONFIG_RCU_EXP_KTHREAD=n. Indeed
synchronize_rcu_expedited() is otherwise implemented on top of a kthread
worker while polling still relies on rcu_gp_wq workqueue, breaking the
above circular dependency chain.
Fix this with making expedited grace period to always rely on kthread
worker. The workqueue based implementation is essentially a duplicate
anyway now that the per-node initialization is performed by per-node
kthread workers.
Meanwhile the CONFIG_RCU_EXP_KTHREAD switch is still kept around to
manage the scheduler policy of these kthread workers.
Reported-by: Anna-Maria Behnsen <anna-maria@linutronix.de>
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Suggested-by: Joel Fernandes <joel@joelfernandes.org>
Suggested-by: Paul E. McKenney <paulmck@kernel.org>
Suggested-by: Neeraj upadhyay <Neeraj.Upadhyay@amd.com>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Affine the parallel expedited gp kworkers to their respective RCU node
in order to make them close to the cache their are playing with.
This reuses the boost kthreads machinery that probe into CPU hotplug
operations such that the kthreads become/stay affine to their respective
node as soon/long as they contain online CPUs. Otherwise and if the
current CPU going down was the last online on the leaf node, the related
kthread is affine to the housekeeping CPUs.
In the long run, this affinity VS CPU hotplug operation game should
probably be implemented at the generic kthread level.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
[boqun: s/* rcu_boost_task/*rcu_boost_task as reported by checkpatch]
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
When CONFIG_RCU_EXP_KTHREAD=n, the expedited grace period per node
initialization is performed in parallel via workqueues (one work per
node).
However in CONFIG_RCU_EXP_KTHREAD=y, this per node initialization is
performed by a single kworker serializing each node initialization (one
work for all nodes).
The second part is certainly less scalable and efficient beyond a single
leaf node.
To improve this, expand this single kworker into per-node kworkers. This
new layout is eventually intended to remove the workqueues based
implementation since it will essentially now become duplicate code.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
The expedited kthread worker performing the per node initialization is
going to be split into per node kthreads. As such, the future per node
kthread creation will need to be called from CPU hotplug callbacks
instead of an initcall, right beside the per node boost kthread
creation.
To prepare for that, move the kthread worker creation above
rcutree_prepare_cpu() as a first step to make the review smoother for
the upcoming modifications.
No intended functional change.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
This mutex is currently protecting per node boost kthreads creation and
affinity setting across CPU hotplug operations.
Since the expedited kworkers will soon be split per node as well, they
will be subject to the same concurrency constraints against hotplug.
Therefore their creation and affinity tuning operations will be grouped
with those of boost kthreads and then rely on the same mutex.
To prepare for that, generalize its name.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Just like is done for the kworker performing nodes initialization,
gracefully handle the possible allocation failure of the RCU expedited
grace period main kworker.
While at it perform a rename of the related checking functions to better
reflect the expedited specifics.
Reviewed-by: Kalesh Singh <kaleshsingh@google.com>
Fixes: 9621fbee44df ("rcu: Move expedited grace period (GP) work to RT kthread_worker")
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Under CONFIG_RCU_EXP_KTHREAD=y, the nodes initialization for expedited
grace periods is queued to a kworker. However if the allocation of that
kworker failed, the nodes initialization is performed synchronously by
the caller instead.
Now the check for kworker initialization failure relies on the kworker
pointer to be NULL while its value might actually encapsulate an
allocation failure error.
Make sure to handle this case.
Reviewed-by: Kalesh Singh <kaleshsingh@google.com>
Fixes: 9621fbee44df ("rcu: Move expedited grace period (GP) work to RT kthread_worker")
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
When an expedited grace period is ending, care must be taken so that all
the quiescent states propagated up to the root are correctly ordered
against the wake up of the main expedited grace period workqueue.
This ordering is already carried through the root rnp locking augmented
by an smp_mb__after_unlock_lock() barrier.
Therefore the explicit smp_mb() placed before the wake up is not needed
and can be removed.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Currently, only rdp_gp->nocb_timer is used, for nocb_timer of
no-rdp_gp structure, the timer_pending() is always return false,
this commit therefore need to check rdp_gp->nocb_timer in
__call_rcu_nocb_wake().
Signed-off-by: Zqiang <qiang.zhang1211@gmail.com>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
For the kernels built with CONFIG_RCU_NOCB_CPU_DEFAULT_ALL=y and
CONFIG_RCU_LAZY=y, the following scenarios will trigger WARN_ON_ONCE()
in the rcu_nocb_bypass_lock() and rcu_nocb_wait_contended() functions:
CPU2 CPU11
kthread
rcu_nocb_cb_kthread ksys_write
rcu_do_batch vfs_write
rcu_torture_timer_cb proc_sys_write
__kmem_cache_free proc_sys_call_handler
kmemleak_free drop_caches_sysctl_handler
delete_object_full drop_slab
__delete_object shrink_slab
put_object lazy_rcu_shrink_scan
call_rcu rcu_nocb_flush_bypass
__call_rcu_commn rcu_nocb_bypass_lock
raw_spin_trylock(&rdp->nocb_bypass_lock) fail
atomic_inc(&rdp->nocb_lock_contended);
rcu_nocb_wait_contended WARN_ON_ONCE(smp_processor_id() != rdp->cpu);
WARN_ON_ONCE(atomic_read(&rdp->nocb_lock_contended)) |
|_ _ _ _ _ _ _ _ _ _same rdp and rdp->cpu != 11_ _ _ _ _ _ _ _ _ __|
Reproduce this bug with "echo 3 > /proc/sys/vm/drop_caches".
This commit therefore uses rcu_nocb_try_flush_bypass() instead of
rcu_nocb_flush_bypass() in lazy_rcu_shrink_scan(). If the nocb_bypass
queue is being flushed, then rcu_nocb_try_flush_bypass will return
directly.
Signed-off-by: Zqiang <qiang.zhang1211@gmail.com>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Currently the call_rcu() function interleaves NOCB and !NOCB enqueue
code in a complicated way such that:
* The bypass enqueue code may or may not have enqueued and may or may
not have locked the ->nocb_lock. Everything that follows is in a
Schrödinger locking state for the unwary reviewer's eyes.
* The was_alldone is always set but only used in NOCB related code.
* The NOCB wake up is distantly related to the locking hopefully
performed by the bypass enqueue code that did not enqueue on the
bypass list.
Unconfuse the whole and gather NOCB and !NOCB specific enqueue code to
their own functions.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Currently IRQs are disabled on call_rcu() and then depending on the
context:
* If the CPU is in nocb mode:
- If the callback is enqueued in the bypass list, IRQs are re-enabled
implictly by rcu_nocb_try_bypass()
- If the callback is enqueued in the normal list, IRQs are re-enabled
implicitly by __call_rcu_nocb_wake()
* If the CPU is NOT in nocb mode, IRQs are reenabled explicitly from call_rcu()
This makes the code a bit hard to follow, especially as it interleaves
with nocb locking.
To make the IRQ flags coverage clearer and also in order to prepare for
moving all the nocb enqueue code to its own function, always re-enable
the IRQ flags explicitly from call_rcu().
Reviewed-by: Neeraj Upadhyay (AMD) <neeraj.iitr10@gmail.com>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
A full barrier is issued from nocb_gp_wait() upon callbacks advancing
to order grace period completion with callbacks execution.
However these two events are already ordered by the
smp_mb__after_unlock_lock() barrier within the call to
raw_spin_lock_rcu_node() that is necessary for callbacks advancing to
happen.
The following litmus test shows the kind of guarantee that this barrier
provides:
C smp_mb__after_unlock_lock
{}
// rcu_gp_cleanup()
P0(spinlock_t *rnp_lock, int *gpnum)
{
// Grace period cleanup increase gp sequence number
spin_lock(rnp_lock);
WRITE_ONCE(*gpnum, 1);
spin_unlock(rnp_lock);
}
// nocb_gp_wait()
P1(spinlock_t *rnp_lock, spinlock_t *nocb_lock, int *gpnum, int *cb_ready)
{
int r1;
// Call rcu_advance_cbs() from nocb_gp_wait()
spin_lock(nocb_lock);
spin_lock(rnp_lock);
smp_mb__after_unlock_lock();
r1 = READ_ONCE(*gpnum);
WRITE_ONCE(*cb_ready, 1);
spin_unlock(rnp_lock);
spin_unlock(nocb_lock);
}
// nocb_cb_wait()
P2(spinlock_t *nocb_lock, int *cb_ready, int *cb_executed)
{
int r2;
// rcu_do_batch() -> rcu_segcblist_extract_done_cbs()
spin_lock(nocb_lock);
r2 = READ_ONCE(*cb_ready);
spin_unlock(nocb_lock);
// Actual callback execution
WRITE_ONCE(*cb_executed, 1);
}
P3(int *cb_executed, int *gpnum)
{
int r3;
WRITE_ONCE(*cb_executed, 2);
smp_mb();
r3 = READ_ONCE(*gpnum);
}
exists (1:r1=1 /\ 2:r2=1 /\ cb_executed=2 /\ 3:r3=0) (* Bad outcome. *)
Here the bad outcome only occurs if the smp_mb__after_unlock_lock() is
removed. This barrier orders the grace period completion against
callbacks advancing and even later callbacks invocation, thanks to the
opportunistic propagation via the ->nocb_lock to nocb_cb_wait().
Therefore the smp_mb() placed after callbacks advancing can be safely
removed.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
The LOAD-ACQUIRE access performed on rdp->nocb_cb_sleep advertizes
ordering callback execution against grace period completion. However
this is contradicted by the following:
* This LOAD-ACQUIRE doesn't pair with anything. The only counterpart
barrier that can be found is the smp_mb() placed after callbacks
advancing in nocb_gp_wait(). However the barrier is placed _after_
->nocb_cb_sleep write.
* Callbacks can be concurrently advanced between the LOAD-ACQUIRE on
->nocb_cb_sleep and the call to rcu_segcblist_extract_done_cbs() in
rcu_do_batch(), making any ordering based on ->nocb_cb_sleep broken.
* Both rcu_segcblist_extract_done_cbs() and rcu_advance_cbs() are called
under the nocb_lock, the latter hereby providing already the desired
ACQUIRE semantics.
Therefore it is safe to access ->nocb_cb_sleep with a simple compiler
barrier.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
When the CPU goes idle for the last time during the CPU down hotplug
process, RCU reports a final quiescent state for the current CPU. If
this quiescent state propagates up to the top, some tasks may then be
woken up to complete the grace period: the main grace period kthread
and/or the expedited main workqueue (or kworker).
If those kthreads have a SCHED_FIFO policy, the wake up can indirectly
arm the RT bandwith timer to the local offline CPU. Since this happens
after hrtimers have been migrated at CPUHP_AP_HRTIMERS_DYING stage, the
timer gets ignored. Therefore if the RCU kthreads are waiting for RT
bandwidth to be available, they may never be actually scheduled.
This triggers TREE03 rcutorture hangs:
rcu: INFO: rcu_preempt self-detected stall on CPU
rcu: 4-...!: (1 GPs behind) idle=9874/1/0x4000000000000000 softirq=0/0 fqs=20 rcuc=21071 jiffies(starved)
rcu: (t=21035 jiffies g=938281 q=40787 ncpus=6)
rcu: rcu_preempt kthread starved for 20964 jiffies! g938281 f0x0 RCU_GP_WAIT_FQS(5) ->state=0x0 ->cpu=0
rcu: Unless rcu_preempt kthread gets sufficient CPU time, OOM is now expected behavior.
rcu: RCU grace-period kthread stack dump:
task:rcu_preempt state:R running task stack:14896 pid:14 tgid:14 ppid:2 flags:0x00004000
Call Trace:
<TASK>
__schedule+0x2eb/0xa80
schedule+0x1f/0x90
schedule_timeout+0x163/0x270
? __pfx_process_timeout+0x10/0x10
rcu_gp_fqs_loop+0x37c/0x5b0
? __pfx_rcu_gp_kthread+0x10/0x10
rcu_gp_kthread+0x17c/0x200
kthread+0xde/0x110
? __pfx_kthread+0x10/0x10
ret_from_fork+0x2b/0x40
? __pfx_kthread+0x10/0x10
ret_from_fork_asm+0x1b/0x30
</TASK>
The situation can't be solved with just unpinning the timer. The hrtimer
infrastructure and the nohz heuristics involved in finding the best
remote target for an unpinned timer would then also need to handle
enqueues from an offline CPU in the most horrendous way.
So fix this on the RCU side instead and defer the wake up to an online
CPU if it's too late for the local one.
Reported-by: Paul E. McKenney <paulmck@kernel.org>
Fixes: 5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU earlier")
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Neeraj Upadhyay (AMD) <neeraj.iitr10@gmail.com>
If an rcutorture test scenario creates an fqs_task kthread, it will
periodically invoke rcu_force_quiescent_state() in order to start
force-quiescent-state (FQS) operations. However, an FQS operation
will be started even if there is no RCU grace period in progress.
Although testing FQS operations startup when there is no grace period in
progress is necessary, it need not happen all that often. This commit
therefore causes rcu_force_quiescent_state() to take an early exit
if there is no grace period in progress.
Note that there will still be attempts to start an FQS scan in the
absence of a grace period because the grace period might end right
after the rcu_force_quiescent_state() function's check. In actual
testing, this happens about once every ten minutes, which should
provide adequate testing.
Signed-off-by: Zqiang <qiang.zhang1211@gmail.com>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Neeraj Upadhyay (AMD) <neeraj.iitr10@gmail.com>
If an SRCU barrier is queued while callbacks are running and a new
callbacks invocator for the same sdp were to run concurrently, the
RCU barrier might execute too early. As this requirement is non-obvious,
make sure to keep a record.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Neeraj Upadhyay (AMD) <neeraj.iitr10@gmail.com>
While in grace period start, there is nothing to accelerate and
therefore no need to advance the callbacks either if no callback is
to be enqueued.
Spare these needless operations in this case.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Neeraj Upadhyay (AMD) <neeraj.iitr10@gmail.com>
Callbacks advancing on SRCU must be performed on two specific places:
1) On enqueue time in order to make room for the acceleration of the
new callback.
2) On invocation time in order to move the callbacks ready to invoke.
Any other callback advancing callsite is needless. Remove the remaining
one in srcu_gp_start().
Co-developed-by: Yong He <zhuangel570@gmail.com>
Signed-off-by: Yong He <zhuangel570@gmail.com>
Co-developed-by: Joel Fernandes <joel@joelfernandes.org>
Signed-off-by: Joel Fernandes <joel@joelfernandes.org>
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Co-developed-by: Neeraj Upadhyay (AMD) <neeraj.iitr10@gmail.com>
Signed-off-by: Neeraj Upadhyay (AMD) <neeraj.iitr10@gmail.com>
