Current rcuperf startup checks to see if the user asked to measure
only expedited grace periods, yet constrained all grace periods to be
normal, or if the user asked to measure only normal grace periods, yet
constrained all grace periods to be expedited. Useless tests of this
sort are aborted.
Unfortunately, making RCU work through the mid-boot dead zone [1] puts
RCU into expedited-only mode during that zone. Which happens to also
be the exact time that rcuperf carries out the aforementioned check.
So if the user asks rcuperf to measure only normal grace periods (the
default), rcuperf will now always complain and terminate the test.
This commit therefore moves the checks to rcu_perf_cleanup(). This has
the disadvantage of failing to abort useless tests, but avoids the need to
create yet another kthread and the need to do fiddly checks involving the
holdoff time. (Yes, another approach is to do the checks in a late-stage
init function, but that would require some way to communicate badness
to rcuperf's kthreads, and seems not worth the bother.)
[1] https://lwn.net/Articles/716148/
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Although preemptible RCU allows its read-side critical sections to be
preempted, general blocking is forbidden. The reason for this is that
excessive preemption times can be handled by CONFIG_RCU_BOOST=y, but a
voluntarily blocked task doesn't care how high you boost its priority.
Because preemptible RCU is a global mechanism, one ill-behaved reader
hurts everyone. Hence the prohibition against general blocking in
RCU-preempt read-side critical sections. Preemption yes, blocking no.
This commit enforces this prohibition.
There is a special exception for the -rt patchset (which they kindly
volunteered to implement): It is OK to block (as opposed to merely being
preempted) within an RCU-preempt read-side critical section, but only if
the blocking is subject to priority inheritance. This exception permits
CONFIG_RCU_BOOST=y to get -rt RCU readers out of trouble.
Why doesn't this exception also apply to mainline's rt_mutex? Because
of the possibility that someone does general blocking while holding
an rt_mutex. Yes, the priority boosting will affect the rt_mutex,
but it won't help with the task doing general blocking while holding
that rt_mutex.
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Earlier versions of Tree SRCU were subject to a counter overflow bug that
could theoretically result in too-short grace periods. This commit
eliminates this problem by adding an update-side memory barrier.
The short explanation is that if the updater sums the unlock counts
too late to see a given __srcu_read_unlock() increment, that CPU's
next __srcu_read_lock() must see the new value of ->srcu_idx, thus
incrementing the other bank of counters. This eliminates the possibility
of destructive counter overflow as long as the srcu_read_lock() nesting
level does not exceed floor(ULONG_MAX/NR_CPUS/2), which should be an
eminently reasonable nesting limit, especially on 64-bit systems.
Reported-by: Lance Roy <ldr709@gmail.com>
Suggested-by: Lance Roy <ldr709@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Currently rcu_barrier() uses call_rcu() to enqueue new callbacks
on each CPU with a non-empty callback list. This works, but means
that rcu_barrier() forces grace periods that are not otherwise needed.
The key point is that rcu_barrier() never needs to wait for a grace
period, but instead only for all pre-existing callbacks to be invoked.
This means that rcu_barrier()'s new callbacks should be placed in
the callback-list segment containing the last pre-existing callback.
This commit makes this change using the new rcu_segcblist_entrain()
function.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Linu Cherian reported a WARN in cleanup_srcu_struct() when shutting
down a guest running iperf on a VFIO assigned device. This happens
because irqfd_wakeup() calls srcu_read_lock(&kvm->irq_srcu) in interrupt
context, while a worker thread does the same inside kvm_set_irq(). If the
interrupt happens while the worker thread is executing __srcu_read_lock(),
updates to the Classic SRCU ->lock_count[] field or the Tree SRCU
->srcu_lock_count[] field can be lost.
The docs say you are not supposed to call srcu_read_lock() and
srcu_read_unlock() from irq context, but KVM interrupt injection happens
from (host) interrupt context and it would be nice if SRCU supported the
use case. KVM is using SRCU here not really for the "sleepable" part,
but rather due to its IPI-free fast detection of grace periods. It is
therefore not desirable to switch back to RCU, which would effectively
revert commit 719d93cd5f ("kvm/irqchip: Speed up KVM_SET_GSI_ROUTING",
2014-01-16).
However, the docs are overly conservative. You can have an SRCU instance
only has users in irq context, and you can mix process and irq context
as long as process context users disable interrupts. In addition,
__srcu_read_unlock() actually uses this_cpu_dec() on both Tree SRCU and
Classic SRCU. For those two implementations, only srcu_read_lock()
is unsafe.
When Classic SRCU's __srcu_read_unlock() was changed to use this_cpu_dec(),
in commit 5a41344a3d ("srcu: Simplify __srcu_read_unlock() via
this_cpu_dec()", 2012-11-29), __srcu_read_lock() did two increments.
Therefore it kept __this_cpu_inc(), with preempt_disable/enable in
the caller. Tree SRCU however only does one increment, so on most
architectures it is more efficient for __srcu_read_lock() to use
this_cpu_inc(), and any performance differences appear to be down in
the noise.
Cc: stable@vger.kernel.org
Fixes: 719d93cd5f ("kvm/irqchip: Speed up KVM_SET_GSI_ROUTING")
Reported-by: Linu Cherian <linuc.decode@gmail.com>
Suggested-by: Linu Cherian <linuc.decode@gmail.com>
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Linu Cherian reported a WARN in cleanup_srcu_struct() when shutting
down a guest running iperf on a VFIO assigned device. This happens
because irqfd_wakeup() calls srcu_read_lock(&kvm->irq_srcu) in interrupt
context, while a worker thread does the same inside kvm_set_irq(). If the
interrupt happens while the worker thread is executing __srcu_read_lock(),
updates to the Classic SRCU ->lock_count[] field or the Tree SRCU
->srcu_lock_count[] field can be lost.
The docs say you are not supposed to call srcu_read_lock() and
srcu_read_unlock() from irq context, but KVM interrupt injection happens
from (host) interrupt context and it would be nice if SRCU supported the
use case. KVM is using SRCU here not really for the "sleepable" part,
but rather due to its IPI-free fast detection of grace periods. It is
therefore not desirable to switch back to RCU, which would effectively
revert commit 719d93cd5f ("kvm/irqchip: Speed up KVM_SET_GSI_ROUTING",
2014-01-16).
However, the docs are overly conservative. You can have an SRCU instance
only has users in irq context, and you can mix process and irq context
as long as process context users disable interrupts. In addition,
__srcu_read_unlock() actually uses this_cpu_dec() on both Tree SRCU and
Classic SRCU. For those two implementations, only srcu_read_lock()
is unsafe.
When Classic SRCU's __srcu_read_unlock() was changed to use this_cpu_dec(),
in commit 5a41344a3d ("srcu: Simplify __srcu_read_unlock() via
this_cpu_dec()", 2012-11-29), __srcu_read_lock() did two increments.
Therefore it kept __this_cpu_inc(), with preempt_disable/enable in
the caller. Tree SRCU however only does one increment, so on most
architectures it is more efficient for __srcu_read_lock() to use
this_cpu_inc(), and any performance differences appear to be down in
the noise.
Unlike Classic and Tree SRCU, Tiny SRCU does increments and decrements on
a single variable. Therefore, as Peter Zijlstra pointed out, Tiny SRCU's
implementation already supports mixed-context use of srcu_read_lock()
and srcu_read_unlock(), at least as long as uses of srcu_read_lock()
and srcu_read_unlock() in each handler are nested and paired properly.
In other words, it is still illegal to (say) invoke srcu_read_lock()
in an interrupt handler and to invoke the matching srcu_read_unlock()
in a softirq handler. Therefore, the only change required for Tiny SRCU
is to its comments.
Fixes: 719d93cd5f ("kvm/irqchip: Speed up KVM_SET_GSI_ROUTING")
Reported-by: Linu Cherian <linuc.decode@gmail.com>
Suggested-by: Linu Cherian <linuc.decode@gmail.com>
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Paolo Bonzini <pbonzini@redhat.com>
Pull RCU updates from Ingo Molnar:
"The main changes are:
- Debloat RCU headers
- Parallelize SRCU callback handling (plus overlapping patches)
- Improve the performance of Tree SRCU on a CPU-hotplug stress test
- Documentation updates
- Miscellaneous fixes"
* 'core-rcu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (74 commits)
rcu: Open-code the rcu_cblist_n_lazy_cbs() function
rcu: Open-code the rcu_cblist_n_cbs() function
rcu: Open-code the rcu_cblist_empty() function
rcu: Separately compile large rcu_segcblist functions
srcu: Debloat the <linux/rcu_segcblist.h> header
srcu: Adjust default auto-expediting holdoff
srcu: Specify auto-expedite holdoff time
srcu: Expedite first synchronize_srcu() when idle
srcu: Expedited grace periods with reduced memory contention
srcu: Make rcutorture writer stalls print SRCU GP state
srcu: Exact tracking of srcu_data structures containing callbacks
srcu: Make SRCU be built by default
srcu: Fix Kconfig botch when SRCU not selected
rcu: Make non-preemptive schedule be Tasks RCU quiescent state
srcu: Expedite srcu_schedule_cbs_snp() callback invocation
srcu: Parallelize callback handling
kvm: Move srcu_struct fields to end of struct kvm
rcu: Fix typo in PER_RCU_NODE_PERIOD header comment
rcu: Use true/false in assignment to bool
rcu: Use bool value directly
...
Because the rcu_cblist_n_lazy_cbs() just samples the ->len_lazy counter,
and because the rcu_cblist structure is quite straightforward, it makes
sense to open-code rcu_cblist_n_lazy_cbs(p) as p->len_lazy, cutting out
a level of indirection. This commit makes this change.
Reported-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Because the rcu_cblist_n_cbs() just samples the ->len counter, and
because the rcu_cblist structure is quite straightforward, it makes
sense to open-code rcu_cblist_n_cbs(p) as p->len, cutting out a level
of indirection. This commit makes this change.
Reported-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Because the rcu_cblist_empty() just samples the ->head pointer, and
because the rcu_cblist structure is quite straightforward, it makes
sense to open-code rcu_cblist_empty(p) as !p->head, cutting out a
level of indirection. This commit makes this change.
Reported-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
This commit creates a new kernel/rcu/rcu_segcblist.c file that
contains non-trivial segcblist functions. Trivial functions
remain as static inline functions in kernel/rcu/rcu_segcblist.h
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Linus noticed that the <linux/rcu_segcblist.h> has huge inline functions
which should not be inline at all.
As a first step in cleaning this up, move them all to kernel/rcu/ and
only keep an absolute minimum of data type defines in the header:
before: -rw-r--r-- 1 mingo mingo 22284 May 2 10:25 include/linux/rcu_segcblist.h
after: -rw-r--r-- 1 mingo mingo 3180 May 2 10:22 include/linux/rcu_segcblist.h
More can be done, such as uninlining the large functions, which inlining
is unjustified even if it's an RCU internal matter.
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The default value for the kernel boot parameter srcutree.exp_holdoff
is 50 microseconds, which is too long for good Tree SRCU performance
(compared to Classic SRCU) on the workloads tested by Mike Galbraith.
This commit therefore sets the default value to 25 microseconds, which
shows excellent results in Mike's testing.
Reported-by: Mike Galbraith <efault@gmx.de>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Mike Galbraith <efault@gmx.de>
On small systems, in the absence of readers, expedited SRCU grace
periods can complete in less than a microsecond. This means that an
eight-CPU system can have all CPUs doing synchronize_srcu() in a tight
loop and almost always expedite. This might actually be desirable in
some situations, but in general it is a good way to needlessly burn
CPU cycles. And in those situations where it is desirable, your friend
is the function synchronize_srcu_expedited().
For other situations, this commit adds a kernel parameter that specifies
a holdoff between completing the last SRCU grace period and auto-expediting
the next. If the next grace period starts before the holdoff expires,
auto-expediting is disabled. The holdoff is 50 microseconds by default,
and can be tuned to the desired number of nanoseconds. A value of zero
disables auto-expediting.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Mike Galbraith <efault@gmx.de>
Classic SRCU in effect expedites the first synchronize_srcu() when SRCU
is idle, and Mike Galbraith demonstrated that some use cases do in fact
rely on this behavior. In particular, Mike showed that Steven Rostedt's
hotplug stress script takes 55 seconds with Classic SRCU and more than
16 -minutes- when running Tree SRCU. Assuming that each Tree SRCU's call
to synchronize_srcu() takes four milliseconds, this implies that Steven's
test invokes synchronize_srcu() in isolation, but more than once per
200 microseconds. Mike used ftrace to demonstrate that the time between
successive calls to synchronize_srcu() ranged from 118 to 342 microseconds,
with one outlier at 80 milliseconds. This data clearly indicates that
Tree SRCU needs to expedite the first invocation of synchronize_srcu()
during an SRCU idle period.
This commit therefor introduces a srcu_might_be_idle() function that
probabilistically checks whether or not SRCU is idle. This function is
used by synchronize_rcu() as an additional criterion in deciding whether
or not to expedite.
(Hat trick to Peter Zijlstra for his earlier suggestion that this might
in fact be a problem. Which for all I know might have motivated Mike to
look into it.)
Reported-by: Mike Galbraith <efault@gmx.de>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Mike Galbraith <efault@gmx.de>
Commit f60d231a87 ("srcu: Crude control of expedited grace periods")
introduced a per-srcu_struct atomic counter to track outstanding
requests for grace periods. This works, but represents a memory-contention
bottleneck. This commit therefore uses the srcu_node combining tree
to remove this bottleneck.
This commit adds new ->srcu_gp_seq_needed_exp fields to the
srcu_data, srcu_node, and srcu_struct structures, which track the
farthest-in-the-future grace period that must be expedited, which in
turn requires that all nearer-term grace periods also be expedited.
Requests for expediting start with the srcu_data structure, run up
through the srcu_node tree, and end at the srcu_struct structure.
Note that it may be necessary to expedite a grace period that just
now started, and this is handled by a new srcu_funnel_exp_start()
function, which is invoked when the grace period itself is already
in its way, but when that grace period was not marked as expedited.
A new srcu_get_delay() function returns zero if there is at least one
expedited SRCU grace period in flight, or SRCU_INTERVAL otherwise.
This function is used to calculate delays: Normal grace periods
are allowed to extend in order to cover more requests with a given
grace-period computation, which decreases per-request overhead.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Mike Galbraith <efault@gmx.de>
In the past, SRCU was simple enough that there was little point in
making the rcutorture writer stall messages print the SRCU grace-period
number state. With the advent of Tree SRCU, this has changed. This
commit therefore makes Classic, Tiny, and Tree SRCU report this state
to rcutorture as needed.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Mike Galbraith <efault@gmx.de>
The current Tree SRCU implementation schedules a workqueue for every
srcu_data covered by a given leaf srcu_node structure having callbacks,
even if only one of those srcu_data structures actually contains
callbacks. This is clearly inefficient for workloads that don't feature
callbacks everywhere all the time. This commit therefore adds an array
of masks that are used by the leaf srcu_node structures to track exactly
which srcu_data structures contain callbacks.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Mike Galbraith <efault@gmx.de>
Currently, a call to schedule() acts as a Tasks RCU quiescent state
only if a context switch actually takes place. However, just the
call to schedule() guarantees that the calling task has moved off of
whatever tracing trampoline that it might have been one previously.
This commit therefore plumbs schedule()'s "preempt" parameter into
rcu_note_context_switch(), which then records the Tasks RCU quiescent
state, but only if this call to schedule() was -not- due to a preemption.
To avoid adding overhead to the common-case context-switch path,
this commit hides the rcu_note_context_switch() check under an existing
non-common-case check.
Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Although Tree SRCU does reduce delays when there is at least one
synchronize_srcu_expedited() invocation pending, srcu_schedule_cbs_snp()
still waits for SRCU_INTERVAL before invoking callbacks. Since
synchronize_srcu_expedited() now posts a callback and waits for
that callback to do a wakeup, this destroys the expedited nature of
synchronize_srcu_expedited(). This destruction became apparent to
Marc Zyngier in the guise of a guest-OS bootup slowdown from five
seconds to no fewer than forty seconds.
This commit therefore invokes callbacks immediately at the end of the
grace period when there is at least one synchronize_srcu_expedited()
invocation pending. This brought Marc's guest-OS bootup times back
into the realm of reason.
Reported-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Marc Zyngier <marc.zyngier@arm.com>
Peter Zijlstra proposed using SRCU to reduce mmap_sem contention [1,2],
however, there are workloads that could result in a high volume of
concurrent invocations of call_srcu(), which with current SRCU would
result in excessive lock contention on the srcu_struct structure's
->queue_lock, which protects SRCU's callback lists. This commit therefore
moves SRCU to per-CPU callback lists, thus greatly reducing contention.
Because a given SRCU instance no longer has a single centralized callback
list, starting grace periods and invoking callbacks are both more complex
than in the single-list Classic SRCU implementation. Starting grace
periods and handling callbacks are now handled using an srcu_node tree
that is in some ways similar to the rcu_node trees used by RCU-bh,
RCU-preempt, and RCU-sched (for example, the srcu_node tree shape is
controlled by exactly the same Kconfig options and boot parameters that
control the shape of the rcu_node tree).
In addition, the old per-CPU srcu_array structure is now named srcu_data
and contains an rcu_segcblist structure named ->srcu_cblist for its
callbacks (and a spinlock to protect this). The srcu_struct gets
an srcu_gp_seq that is used to associate callback segments with the
corresponding completion-time grace-period number. These completion-time
grace-period numbers are propagated up the srcu_node tree so that the
grace-period workqueue handler can determine whether additional grace
periods are needed on the one hand and where to look for callbacks that
are ready to be invoked.
The srcu_barrier() function must now wait on all instances of the per-CPU
->srcu_cblist. Because each ->srcu_cblist is protected by ->lock,
srcu_barrier() can remotely add the needed callbacks. In theory,
it could also remotely start grace periods, but in practice doing so
is complex and racy. And interestingly enough, it is never necessary
for srcu_barrier() to start a grace period because srcu_barrier() only
enqueues a callback when a callback is already present--and it turns out
that a grace period has to have already been started for this pre-existing
callback. Furthermore, it is only the callback that srcu_barrier()
needs to wait on, not any particular grace period. Therefore, a new
rcu_segcblist_entrain() function enqueues the srcu_barrier() function's
callback into the same segment occupied by the last pre-existing callback
in the list. The special case where all the pre-existing callbacks are
on a different list (because they are in the process of being invoked)
is handled by enqueuing srcu_barrier()'s callback into the RCU_DONE_TAIL
segment, relying on the done-callbacks check that takes place after all
callbacks are inovked.
Note that the readers use the same algorithm as before. Note that there
is a separate srcu_idx that tells the readers what counter to increment.
This unfortunately cannot be combined with srcu_gp_seq because they
need to be incremented at different times.
This commit introduces some ugly #ifdefs in rcutorture. These will go
away when I feel good enough about Tree SRCU to ditch Classic SRCU.
Some crude performance comparisons, courtesy of a quickly hacked rcuperf
asynchronous-grace-period capability:
Callback Queuing Overhead
-------------------------
# CPUS Classic SRCU Tree SRCU
------ ------------ ---------
2 0.349 us 0.342 us
16 31.66 us 0.4 us
41 --------- 0.417 us
The times are the 90th percentiles, a statistic that was chosen to reject
the overheads of the occasional srcu_barrier() call needed to avoid OOMing
the test machine. The rcuperf test hangs when running Classic SRCU at 41
CPUs, hence the line of dashes. Despite the hacks to both the rcuperf code
and that statistics, this is a convincing demonstration of Tree SRCU's
performance and scalability advantages.
[1] https://lwn.net/Articles/309030/
[2] https://patchwork.kernel.org/patch/5108281/
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
[ paulmck: Fix initialization if synchronize_srcu_expedited() called first. ]
This commit just changes a "the the" to "the" to reduce repetition.
Reported-by: Michalis Kokologiannakis <mixaskok@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit makes the parse_rcu_nocb_poll() function assign true
(rather than the constant 1) to the bool variable rcu_nocb_poll.
Signed-off-by: Nicholas Mc Guire <der.herr@hofr.at>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The beenonline variable is declared bool so there is no need for an
explicit comparison, especially not against the constant zero.
Signed-off-by: Nicholas Mc Guire <der.herr@hofr.at>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The rcu_nocb_gp_cleanup() function is now invoked elsewhere, so this
commit drags this comment into the year 2017.
Reported-by: Michalis Kokologiannakis <mixaskok@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The TREE_SRCU rewrite is large and a bit on the non-simple side, so
this commit helps reduce risk by allowing the old v4.11 SRCU algorithm
to be selected using a new CLASSIC_SRCU Kconfig option that depends
on RCU_EXPERT. The default is to use the new TREE_SRCU and TINY_SRCU
algorithms, in order to help get these the testing that they need.
However, if your users do not require the update-side scalability that
is to be provided by TREE_SRCU, select RCU_EXPERT and then CLASSIC_SRCU
to revert back to the old classic SRCU algorithm.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The srcu_torture_stats() function is adapted to the specific srcu_struct
layout traditionally used by SRCU. This commit therefore adds support
for Tiny SRCU.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
In response to automated complaints about modifications to SRCU
increasing its size, this commit creates a tiny SRCU that is
used in SMP=n && PREEMPT=n builds.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
SRCU's implementation of expedited grace periods has always assumed
that the SRCU instance is idle when the expedited request arrives.
This commit improves this a bit by maintaining a count of the number
of outstanding expedited requests, thus allowing prior non-expedited
grace periods accommodate these requests by shifting to expedited mode.
However, any non-expedited wait already in progress will still wait for
the full duration.
Improved control of expedited grace periods is planned, but one step
at a time.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Updating ->srcu_state and ->srcu_gp_seq will lead to extremely complex
race conditions given multiple callback queues, so this commit takes
advantage of the two-bit state now available in rcu_seq counters to
store the state in the bottom two bits of ->srcu_gp_seq.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit increases the number of reserved bits at the bottom of an
rcu_seq grace-period counter from one to two, as will be needed to
accommodate SRCU's three-state grace periods.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The expedited grace-period code contains several open-coded shifts
know the format of an rcu_seq grace-period counter, which is not
particularly good style. This commit therefore creates a new
rcu_seq_ctr() function that extracts the counter portion of the
counter, and an rcu_seq_state() function that extracts the low-order
state bit. This commit prepares for SRCU callback parallelization,
which will require two state bits.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit makes the num_rcu_lvl[] array external so that SRCU can
make use of it for initializing its upcoming srcu_node tree.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit moves rcu_for_each_node_breadth_first(),
rcu_for_each_nonleaf_node_breadth_first(), and
rcu_for_each_leaf_node() from kernel/rcu/tree.h to
kernel/rcu/rcu.h so that SRCU can access them.
This commit is code-movement only.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit moves the rcu_init_levelspread() function from
kernel/rcu/tree.c to kernel/rcu/rcu.h so that SRCU can access it. This is
another step towards enabling SRCU to create its own combining tree.
This commit is code-movement only, give or take knock-on adjustments.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit moves the C preprocessor code that defines the default shape
of the rcu_node combining tree to a new include/linux/rcu_node_tree.h
file as a first step towards enabling SRCU to create its own combining
tree, which in turn enables SRCU to implement per-CPU callback handling,
thus avoiding contention on the lock currently guarding the single list
of callbacks. Note that users of SRCU still need to know the size of
the srcu_struct structure, hence include/linux rather than kernel/rcu.
This commit is code-movement only.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit switches SRCU from custom-built callback queues to the new
rcu_segcblist structure. This change associates grace-period sequence
numbers with groups of callbacks, which will be needed for efficient
processing of per-CPU callbacks.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit adds grace-period sequence numbers, which will be used to
handle mid-boot grace periods and per-CPU callback lists.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The current SRCU grace-period processing might never reach the last
portion of srcu_advance_batches(). This is OK given the current
implementation, as the first portion, up to the try_check_zero()
following the srcu_flip() is sufficient to drive grace periods forward.
However, it has the unfortunate side-effect of making it impossible to
determine when a given grace period has ended, and it will be necessary
to efficiently trace ends of grace periods in order to efficiently handle
per-CPU SRCU callback lists.
This commit therefore adds states to the SRCU grace-period processing,
so that the end of a given SRCU grace period is marked by the transition
to the SRCU_STATE_DONE state.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit simplifies the SRCU state machine by pushing the
srcu_advance_batches() idle-SRCU fastpath into the common case. This is
done by giving srcu_reschedule() a delay parameter, which is zero in
the call from srcu_advance_batches().
This commit is a step towards numbering callbacks in order to
efficiently handle per-CPU callback lists.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The rcu_seq_end() function increments seq signifying completion
of a grace period, after that checks that the seq is even and wakes
_synchronize_rcu_expedited(). The _synchronize_rcu_expedited() function
uses wait_event() to wait for even seq. The problem is that wait_event()
can return as soon as seq becomes even without waiting for the wakeup.
In such case the warning in rcu_seq_end() can falsely fire if the next
expedited grace period starts before the check.
Check that seq has good value before incrementing it.
Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Cc: syzkaller@googlegroups.com
Cc: linux-kernel@vger.kernel.org
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: josh@joshtriplett.org
Cc: jiangshanlai@gmail.com
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
syzkaller-triggered warning:
WARNING: CPU: 0 PID: 4832 at kernel/rcu/tree.c:3533
rcu_seq_end+0x110/0x140 kernel/rcu/tree.c:3533
CPU: 0 PID: 4832 Comm: kworker/0:3 Not tainted 4.10.0+ #276
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
Workqueue: events wait_rcu_exp_gp
Call Trace:
__dump_stack lib/dump_stack.c:15 [inline]
dump_stack+0x2ee/0x3ef lib/dump_stack.c:51
panic+0x1fb/0x412 kernel/panic.c:179
__warn+0x1c4/0x1e0 kernel/panic.c:540
warn_slowpath_null+0x2c/0x40 kernel/panic.c:583
rcu_seq_end+0x110/0x140 kernel/rcu/tree.c:3533
rcu_exp_gp_seq_end kernel/rcu/tree_exp.h:36 [inline]
rcu_exp_wait_wake+0x8a9/0x1330 kernel/rcu/tree_exp.h:517
rcu_exp_sel_wait_wake kernel/rcu/tree_exp.h:559 [inline]
wait_rcu_exp_gp+0x83/0xc0 kernel/rcu/tree_exp.h:570
process_one_work+0xc06/0x1c20 kernel/workqueue.c:2096
worker_thread+0x223/0x19c0 kernel/workqueue.c:2230
kthread+0x326/0x3f0 kernel/kthread.c:227
ret_from_fork+0x31/0x40 arch/x86/entry/entry_64.S:430
---
Expedited grace periods use workqueue handlers that wake up the requesters,
but there is no lock mediating this wakeup. Therefore, memory barriers
are required to ensure that the handler's memory references are seen by
all to occur before synchronize_*_expedited() returns to its caller.
Possibly detected by syzkaller.
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit moves rcu_seq_start(), rcu_seq_end(), rcu_seq_snap(),
and rcu_seq_done() from kernel/rcu/tree.c to kernel/rcu/rcu.h.
This will allow SRCU to use these functions, which in turn will
allow SRCU to move from a single global callback queue to a
per-CPU callback queue.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit adds single-element dequeue functions to rcu_segcblist.
These are less efficient than using the extract and insert functions,
but allow more precise debugging code. These functions are thus
expected to be used only in debug builds, for example, CONFIG_PROVE_RCU.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This commit checks for pre-scheduler state, and if that early in the
boot process, synchronize_srcu() and friends are no-ops.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This is primarily a code-movement commit in preparation for allowing
SRCU to handle early-boot SRCU grace periods.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
RCU has only one multi-tail callback list, which is implemented via
the nxtlist, nxttail, nxtcompleted, qlen_lazy, and qlen fields in the
rcu_data structure, and whose operations are open-code throughout the
Tree RCU implementation. This has been more or less OK in the past,
but upcoming callback-list optimizations in SRCU could really use
a multi-tail callback list there as well.
This commit therefore abstracts the multi-tail callback list handling
into a new kernel/rcu/rcu_segcblist.h file, and uses this new API.
The simple head-and-tail pointer callback list is also abstracted and
applied everywhere except for the NOCB callback-offload lists. (Yes,
the plan is to apply them there as well, but this commit is already
bigger than would be good.)
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
If the RCU_EXPERT Kconfig option is not set (the default), then the
RCU_FANOUT_LEAF Kconfig option will not be defined, which will cause
the leaf-level rcu_node tree fanout to default to 32 on 32-bit systems
and 64 on 64-bit systems. This can result in excessive lock contention.
This commit therefore changes the computation of the leaf-level rcu_node
tree fanout so that the result will be 16 unless an explicit Kconfig or
kernel-boot setting says otherwise.
Reported-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The rcu_all_qs() and rcu_note_context_switch() do a series of checks,
taking various actions to supply RCU with quiescent states, depending
on the outcomes of the various checks. This is a bit much for scheduling
fastpaths, so this commit creates a separate ->rcu_urgent_qs field in
the rcu_dynticks structure that acts as a global guard for these checks.
Thus, in the common case, rcu_all_qs() and rcu_note_context_switch()
check the ->rcu_urgent_qs field, find it false, and simply return.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
The rcu_momentary_dyntick_idle() function scans the RCU flavors, checking
that one of them still needs a quiescent state before doing an expensive
atomic operation on the ->dynticks counter. However, this check reduces
overhead only after a rare race condition, and increases complexity. This
commit therefore removes the scan and the mechanism enabling the scan.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The rcu_qs_ctr variable is yet another isolated per-CPU variable,
so this commit pulls it into the pre-existing rcu_dynticks per-CPU
structure.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The rcu_sched_qs_mask variable is yet another isolated per-CPU variable,
so this commit pulls it into the pre-existing rcu_dynticks per-CPU
structure.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The current use of "RCU_TRACE(statement);" can cause odd bugs, especially
where "statement" is a local-variable declaration, as it can leave a
misplaced ";" in the source code. This commit therefore converts these
to "RCU_TRACE(statement;)", which avoids the misplaced ";".
Reported-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The current use of "RCU_TRACE(statement);" can cause odd bugs, especially
where "statement" is a local-variable declaration, as it can leave a
misplaced ";" in the source code. This commit therefore converts these
to "RCU_TRACE(statement;)", which avoids the misplaced ";".
Reported-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The current use of "RCU_TRACE(statement);" can cause odd bugs, especially
where "statement" is a local-variable declaration, as it can leave a
misplaced ";" in the source code. This commit therefore converts these
to "RCU_TRACE(statement;)", which avoids the misplaced ";".
Reported-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Users of SRCU are obliged to complete all grace-period activity before
invoking cleanup_srcu_struct(). This means that all calls to either
synchronize_srcu() or synchronize_srcu_expedited() must have returned,
and all calls to call_srcu() must have returned, and the last call to
call_srcu() must have been followed by a call to srcu_barrier().
Furthermore, the caller must have done something to prevent any
further calls to synchronize_srcu(), synchronize_srcu_expedited(),
and call_srcu().
Therefore, if there has ever been an invocation of call_srcu() on
the srcu_struct in question, the sequence of events must be as
follows:
1. Prevent any further calls to call_srcu().
2. Wait for any pre-existing call_srcu() invocations to return.
3. Invoke srcu_barrier().
4. It is now safe to invoke cleanup_srcu_struct().
On the other hand, if there has ever been a call to synchronize_srcu()
or synchronize_srcu_expedited(), the sequence of events must be as
follows:
1. Prevent any further calls to synchronize_srcu() or
synchronize_srcu_expedited().
2. Wait for any pre-existing synchronize_srcu() or
synchronize_srcu_expedited() invocations to return.
3. It is now safe to invoke cleanup_srcu_struct().
If there have been calls to all both types of functions (call_srcu()
and either of synchronize_srcu() and synchronize_srcu_expedited()), then
the caller must do the first three steps of the call_srcu() procedure
above and the first two steps of the synchronize_s*() procedure above,
and only then invoke cleanup_srcu_struct().
Note that cleanup_srcu_struct() does some probabilistic checks
for the caller failing to follow these procedures, in which case
cleanup_srcu_struct() does WARN_ON() and avoids freeing the per-CPU
structures associated with the specified srcu_struct structure.
Reported-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
The srcu_reschedule() function invokes rcu_batch_empty() on each of
the four rcu_batch structures in the srcu_struct in question twice.
Given that this check will also be needed in cleanup_srcu_struct(), this
commit consolidates these four checks into a new rcu_all_batches_empty()
function.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Currently, IPIs are used to force other CPUs to invalidate their TLBs
in response to a kernel virtual-memory mapping change. This works, but
degrades both battery lifetime (for idle CPUs) and real-time response
(for nohz_full CPUs), and in addition results in unnecessary IPIs due to
the fact that CPUs executing in usermode are unaffected by stale kernel
mappings. It would be better to cause a CPU executing in usermode to
wait until it is entering kernel mode to do the flush, first to avoid
interrupting usemode tasks and second to handle multiple flush requests
with a single flush in the case of a long-running user task.
This commit therefore reserves a bit at the bottom of the ->dynticks
counter, which is checked upon exit from extended quiescent states.
If it is set, it is cleared and then a new rcu_eqs_special_exit() macro is
invoked, which, if not supplied, is an empty single-pass do-while loop.
If this bottom bit is set on -entry- to an extended quiescent state,
then a WARN_ON_ONCE() triggers.
This bottom bit may be set using a new rcu_eqs_special_set() function,
which returns true if the bit was set, or false if the CPU turned
out to not be in an extended quiescent state. Please note that this
function refuses to set the bit for a non-nohz_full CPU when that CPU
is executing in usermode because usermode execution is tracked by RCU
as a dyntick-idle extended quiescent state only for nohz_full CPUs.
Reported-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Tracing uses rcu_irq_enter() as a way to make sure that RCU is watching when
it needs to use rcu_read_lock() and friends. This is because tracing can
happen as RCU is about to enter user space, or about to go idle, and RCU
does not watch for RCU read side critical sections as it makes the
transition.
There is a small location within the RCU infrastructure that rcu_irq_enter()
itself will not work. If tracing were to occur in that section it will break
if it tries to use rcu_irq_enter().
Originally, this happens with the stack_tracer, because it will call
save_stack_trace when it encounters stack usage that is greater than any
stack usage it had encountered previously. There was a case where that
happened in the RCU section where rcu_irq_enter() did not work, and lockdep
complained loudly about it. To fix it, stack tracing added a call to be
disabled and RCU would disable stack tracing during the critical section
that rcu_irq_enter() was inoperable. This solution worked, but there are
other cases that use rcu_irq_enter() and it would be a good idea to let RCU
give a way to let others know that rcu_irq_enter() will not work. For
example, in trace events.
Another helpful aspect of this change is that it also moves the per cpu
variable called in the RCU critical section into a cache locale along with
other RCU per cpu variables used in that same location.
I'm keeping the stack_trace_disable() code, as that still could be used in
the future by places that really need to disable it. And since it's only a
static inline, it wont take up any kernel text if it is not used.
Link: http://lkml.kernel.org/r/20170405093207.404f8deb@gandalf.local.home
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
The tracing subsystem started using rcu_irq_entry() and rcu_irq_exit()
(with my blessing) to allow the current _rcuidle alternative tracepoint
name to be dispensed with while still maintaining good performance.
Unfortunately, this causes RCU's dyntick-idle entry code's tracing to
appear to RCU like an interrupt that occurs where RCU is not designed
to handle interrupts.
This commit fixes this problem by moving the zeroing of ->dynticks_nesting
after the offending trace_rcu_dyntick() statement, which narrows the
window of vulnerability to a pair of adjacent statements that are now
marked with comments to that effect.
Link: http://lkml.kernel.org/r/20170405093207.404f8deb@gandalf.local.home
Link: http://lkml.kernel.org/r/20170405193928.GM1600@linux.vnet.ibm.com
Reported-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
We are going to split <linux/sched/debug.h> out of <linux/sched.h>, which
will have to be picked up from other headers and a couple of .c files.
Create a trivial placeholder <linux/sched/debug.h> file that just
maps to <linux/sched.h> to make this patch obviously correct and
bisectable.
Include the new header in the files that are going to need it.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Fix up affected files that include this signal functionality via sched.h.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Fix up missing #includes in other places that rely on sched.h doing that for them.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
We are going to split <linux/sched/signal.h> out of <linux/sched.h>, which
will have to be picked up from other headers and a couple of .c files.
Create a trivial placeholder <linux/sched/signal.h> file that just
maps to <linux/sched.h> to make this patch obviously correct and
bisectable.
Include the new header in the files that are going to need it.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
We are going to move scheduler ABI details to <uapi/linux/sched/types.h>,
which will be used from a number of .c files.
Create empty placeholder header that maps to <linux/types.h>.
Include the new header in the files that are going to need it.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
So rcupdate.h is a pretty complex header, in particular it includes
<linux/completion.h> which includes <linux/wait.h> - creating a
dependency that includes <linux/wait.h> in <linux/sched.h>,
which prevents the isolation of <linux/sched.h> from the derived
<linux/wait.h> header.
Solve part of the problem by decoupling rcupdate.h from completions:
this can be done by separating out the rcu_synchronize types and APIs,
and updating their usage sites.
Since this is a mostly RCU-internal types this will not just simplify
<linux/sched.h>'s dependencies, but will make all the hundreds of
.c files that include rcupdate.h but not completions or wait.h build
faster.
( For rcutiny this means that two dependent APIs have to be uninlined,
but that shouldn't be much of a problem as they are rare variants. )
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Because there are no memory barriers between the srcu_flip() ->completed
increment and the summation of the read-side ->unlock_count[] counters,
both the compiler and the CPU can reorder the summation with the
->completed increment. If the updater is preempted long enough during
this process, the read-side counters could overflow, resulting in a
too-short grace period.
This commit therefore adds a memory barrier just after the ->completed
increment, ensuring that if the summation misses an increment of
->unlock_count[] from __srcu_read_unlock(), the next __srcu_read_lock()
will see the new value of ->completed, thus bounding the number of
->unlock_count[] increments that can be missed to NR_CPUS. The actual
overflow computation is more complex due to the possibility of nesting
of __srcu_read_lock().
Reported-by: Lance Roy <ldr709@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
If a process invokes synchronize_srcu(), is delayed just the right amount
of time, and thus does not sleep when waiting for the grace period to
complete, there is no ordering between the end of the grace period and
the code following the synchronize_srcu(). Similarly, there can be a
lack of ordering between the end of the SRCU grace period and callback
invocation.
This commit adds the necessary ordering.
Reported-by: Lance Roy <ldr709@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
[ paulmck: Further smp_mb() adjustment per email with Lance Roy. ]
SRCU uses two per-cpu counters: a nesting counter to count the number of
active critical sections, and a sequence counter to ensure that the nesting
counters don't change while they are being added together in
srcu_readers_active_idx_check().
This patch instead uses per-cpu lock and unlock counters. Because both
counters only increase and srcu_readers_active_idx_check() reads the unlock
counter before the lock counter, this achieves the same end without having
to increment two different counters in srcu_read_lock(). This also saves a
smp_mb() in srcu_readers_active_idx_check().
Possible bug: There is no guarantee that the lock counter won't overflow
during srcu_readers_active_idx_check(), as there are no memory barriers
around srcu_flip() (see comment in srcu_readers_active_idx_check() for
details). However, this problem was already present before this patch.
Suggested-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Lance Roy <ldr709@gmail.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Commit 7ec99de36f ("rcu: Provide exact CPU-online tracking for RCU"),
as its title suggests, got rid of RCU's remaining CPU-hotplug timing
guesswork. This commit therefore removes the one-jiffy kludge that was
used to paper over this guesswork.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Commit 4a81e8328d ("rcu: Reduce overhead of cond_resched() checks
for RCU") moved quiescent-state generation out of cond_resched()
and commit bde6c3aa99 ("rcu: Provide cond_resched_rcu_qs() to force
quiescent states in long loops") introduced cond_resched_rcu_qs(), and
commit 5cd37193ce ("rcu: Make cond_resched_rcu_qs() apply to normal RCU
flavors") introduced the per-CPU rcu_qs_ctr variable, which is frequently
polled by the RCU core state machine.
This frequent polling can increase grace-period rate, which in turn
increases grace-period overhead, which is visible in some benchmarks
(for example, the "open1" benchmark in Anton Blanchard's "will it scale"
suite). This commit therefore reduces the rate at which rcu_qs_ctr
is polled by moving that polling into the force-quiescent-state (FQS)
machinery, and by further polling it only after the grace period has
been in effect for at least jiffies_till_sched_qs jiffies.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
This commit is the fourth step towards full abstraction of all accesses
to the ->dynticks counter, implementing previously open-coded checks and
comparisons in new rcu_dynticks_in_eqs() and rcu_dynticks_in_eqs_since()
functions. This abstraction will ease changes to the ->dynticks counter
operation.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
This commit is the third step towards full abstraction of all accesses
to the ->dynticks counter, implementing the previously open-coded atomic
add of 1 and entry checks in a new rcu_dynticks_eqs_enter() function, and
the same but with exit checks in a new rcu_dynticks_eqs_exit() function.
This abstraction will ease changes to the ->dynticks counter operation.
Note that this commit gets rid of the smp_mb__before_atomic() and the
smp_mb__after_atomic() calls that were previously present. The reason
that this is OK from a memory-ordering perspective is that the atomic
operation is now atomic_add_return(), which, as a value-returning atomic,
guarantees full ordering.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
[ paulmck: Fixed RCU_TRACE() statements added by this commit. ]
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
The non-expedited synchronize_*rcu() primitives have lockdep checks, but
their expedited counterparts lack these checks. This commit therefore
adds these checks to the expedited synchronize_*rcu() primitives.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Expedited grace periods no longer fall back to normal grace periods
in response to lock contention, given that expedited grace periods
now use the rcu_node tree so as to avoid contention. This commit
therfore removes the expedited_normal counter.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
It used to be that the rcuo callback-offload kthreads were spawned
in rcu_organize_nocb_kthreads(), and the comment before the "for"
loop says as much. However, this spawning has long since moved to
the CPU-hotplug code, so this commit fixes this comment.
Reported-by: Michalis Kokologiannakis <mixaskok@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
The rcu_cpu_starting() function uses this_cpu_ptr() to locate the
incoming CPU's rcu_data structure. This works for the boot CPU and for
all CPUs onlined after rcu_init() executes (during very early boot).
Currently, this is the full set of CPUs, so all is well. But if
anyone ever parallelizes boot before rcu_init() time, it will fail.
This commit therefore substitutes the rcu_cpu_starting() function's
this_cpu_pointer() for per_cpu_ptr(), future-proofing the code and
(arguably) improving readability.
This commit inadvertently fixes a latent bug: If there ever had been
more than just the boot CPU online at rcu_init() time, the old code
would not initialize the non-boot CPUs, but rather would repeatedly
initialize the boot CPU.
Reported-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
These functions (rcu_exp_gp_seq_start(), rcu_exp_gp_seq_end(),
rcu_exp_gp_seq_snap(), and rcu_exp_gp_seq_done() seemed too obvious
to comment when written, but not so much when being documented.
This commit therefore adds header comments to each of them.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Chris Friesen notice that rcuc/X kthreads were consuming CPU even on
NOCB CPUs. This makes no sense because the only purpose or these
kthreads is to invoke normal (non-offloaded) callbacks, of which there
will never be any on NOCB CPUs. This problem was due to a bug in
cpu_has_callbacks_ready_to_invoke(), which should have been checking
->nxttail[RCU_NEXT_TAIL] for NULL, but which was instead (incorrectly)
checking ->nxttail[RCU_DONE_TAIL]. Because ->nxttail[RCU_DONE_TAIL] is
never NULL, the only effect is to cause the rcuc/X kthread to execute
when it should not do so.
This commit therefore checks ->nxttail[RCU_NEXT_TAIL], which is NULL
for NOCB CPUs.
Reported-by: Chris Friesen <chris.friesen@windriver.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
This commit is for all intents and purposes a revert of bc1dce514e
("rcu: Don't use NMIs to dump other CPUs' stacks"). The reason to suppose
that this can now safely be reverted is the presence of 42a0bb3f71
("printk/nmi: generic solution for safe printk in NMI"), which is said
to have made NMI-based stack dumps safe.
However, this reversion keeps one nice property of bc1dce514e
("rcu: Don't use NMIs to dump other CPUs' stacks"), namely that
only those CPUs blocking the grace period are dumped. The new
trigger_single_cpu_backtrace() is used to make this happen, as
suggested by Josh Poimboeuf.
Reported-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Reviewed-by: Josh Poimboeuf <jpoimboe@redhat.com>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Commit 4914950aaa ("rcu: Stop treating in-kernel CPU-bound workloads
as errors") added a (relatively) short-timeout call to resched_cpu().
This was inspired by as issue that was fixed by b7e7ade34e ("sched/core:
Fix remote wakeups"). But given that this issue was fixed, it is time
for the current commit to remove this call to resched_cpu().
Reported-by: Byungchul Park <byungchul.park@lge.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
This commit prepares for the removal of short-term CPU kicking (in a
subsequent commit). It does so by starting to invoke resched_cpu()
for each holdout at each force-quiescent-state interval that is more
than halfway through the stall-warning interval.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Since commit 7ec99de36f ("rcu: Provide exact CPU-online tracking for
RCU"), the variable mask in rcu_init_percpu_data is set but no longer
used. Remove it to fix the following warning when building with 'W=1':
kernel/rcu/tree.c: In function ‘rcu_init_percpu_data’:
kernel/rcu/tree.c:3765:16: warning: variable ‘mask’ set but not used [-Wunused-but-set-variable]
Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
The declarations of __rcu_process_callbacks() and rcu_process_callbacks()
are not needed, as the definition of both of these functions appear before
any uses. This commit therefore removes both declarations.
Reported-by: "Ahmed, Iftekhar" <ahmedi@oregonstate.edu>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
The print_other_cpu_stall() function currently unconditionally invokes
rcu_print_detail_task_stall(). This is OK because if there was a stall
sufficient to cause print_other_cpu_stall() to be invoked, that stall
is very likely to persist through the entire print_other_cpu_stall()
execution. However, if the stall did not persist, the variable ndetected
will be zero, and that variable is already tested in an "if" statement.
Therefore, this commit moves the call to rcu_print_detail_task_stall()
under that pre-existing "if" to improve readability, with a very rare
reduction in overhead.
Signed-off-by: Byungchul Park <byungchul.park@lge.com>
[ paulmck: Reworked commit log. ]
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
RCU_EXPEDITE_BOOT should speed up the boot process by enforcing
synchronize_rcu_expedited() instead of synchronize_rcu() during the boot
process. There should be no reason why one does not want this and there
is no need worry about real time latency at this point.
Therefore make it default.
Note that users wishing to avoid expediting entirely, for example when
bringing up new hardware possibly having flaky IPIs, can use the
rcu_normal boot parameter to override boot-time expediting.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
[ paulmck: Reworded commit log. ]
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
This commit is the second step towards full abstraction of all accesses to
the ->dynticks counter, implementing the previously open-coded atomic
add of zero in a new rcu_dynticks_snap() function. This abstraction will
ease changes o the ->dynticks counter operation.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
This commit is the first step towards full abstraction of all accesses to
the ->dynticks counter, implementing the previously open-coded atomic add
of two in a new rcu_dynticks_momentary_idle() function. This abstraction
will ease changes to the ->dynticks counter operation.
Note that this commit gets rid of the smp_mb__before_atomic() and the
smp_mb__after_atomic() calls that were previously present. The reason
that this is OK from a memory-ordering perspective is that the atomic
operation is now atomic_add_return(), which, as a value-returning atomic,
guarantees full ordering.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
The current preemptible RCU implementation goes through three phases
during bootup. In the first phase, there is only one CPU that is running
with preemption disabled, so that a no-op is a synchronous grace period.
In the second mid-boot phase, the scheduler is running, but RCU has
not yet gotten its kthreads spawned (and, for expedited grace periods,
workqueues are not yet running. During this time, any attempt to do
a synchronous grace period will hang the system (or complain bitterly,
depending). In the third and final phase, RCU is fully operational and
everything works normally.
This has been OK for some time, but there has recently been some
synchronous grace periods showing up during the second mid-boot phase.
This code worked "by accident" for awhile, but started failing as soon
as expedited RCU grace periods switched over to workqueues in commit
8b355e3bc1 ("rcu: Drive expedited grace periods from workqueue").
Note that the code was buggy even before this commit, as it was subject
to failure on real-time systems that forced all expedited grace periods
to run as normal grace periods (for example, using the rcu_normal ksysfs
parameter). The callchain from the failure case is as follows:
early_amd_iommu_init()
|-> acpi_put_table(ivrs_base);
|-> acpi_tb_put_table(table_desc);
|-> acpi_tb_invalidate_table(table_desc);
|-> acpi_tb_release_table(...)
|-> acpi_os_unmap_memory
|-> acpi_os_unmap_iomem
|-> acpi_os_map_cleanup
|-> synchronize_rcu_expedited
The kernel showing this callchain was built with CONFIG_PREEMPT_RCU=y,
which caused the code to try using workqueues before they were
initialized, which did not go well.
This commit therefore reworks RCU to permit synchronous grace periods
to proceed during this mid-boot phase. This commit is therefore a
fix to a regression introduced in v4.9, and is therefore being put
forward post-merge-window in v4.10.
This commit sets a flag from the existing rcu_scheduler_starting()
function which causes all synchronous grace periods to take the expedited
path. The expedited path now checks this flag, using the requesting task
to drive the expedited grace period forward during the mid-boot phase.
Finally, this flag is updated by a core_initcall() function named
rcu_exp_runtime_mode(), which causes the runtime codepaths to be used.
Note that this arrangement assumes that tasks are not sent POSIX signals
(or anything similar) from the time that the first task is spawned
through core_initcall() time.
Fixes: 8b355e3bc1 ("rcu: Drive expedited grace periods from workqueue")
Reported-by: "Zheng, Lv" <lv.zheng@intel.com>
Reported-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Stan Kain <stan.kain@gmail.com>
Tested-by: Ivan <waffolz@hotmail.com>
Tested-by: Emanuel Castelo <emanuel.castelo@gmail.com>
Tested-by: Bruno Pesavento <bpesavento@infinito.it>
Tested-by: Borislav Petkov <bp@suse.de>
Tested-by: Frederic Bezies <fredbezies@gmail.com>
Cc: <stable@vger.kernel.org> # 4.9.0-
It is now legal to invoke synchronize_sched() at early boot, which causes
Tiny RCU's synchronize_sched() to emit spurious splats. This commit
therefore removes the cond_resched() from Tiny RCU's synchronize_sched().
Fixes: 8b355e3bc1 ("rcu: Drive expedited grace periods from workqueue")
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: <stable@vger.kernel.org> # 4.9.0-
The current code can result in spurious kicks when there are no grace
periods in progress and no grace-period-related requests. This is
sort of OK for a diagnostic aid, but the resulting ftrace-dump messages
in dmesg are annoying. This commit therefore avoids spurious kicks
in the common case.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Expedited grace periods check dyntick-idle state, and avoid sending
IPIs to idle CPUs, including those running guest OSes, and, on NOHZ_FULL
kernels, nohz_full CPUs. However, the kernel has been observed checking
a CPU while it was non-idle, but sending the IPI after it has gone
idle. This commit therefore rechecks idle state immediately before
sending the IPI, refraining from IPIing CPUs that have since gone idle.
Reported-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Although rcutorture will occasionally do a 50-millisecond grace-period
delay, these delays are quite rare. And rightly so, because otherwise
the read rate would be quite low. Thie means that it can be important
to identify whether or not a given run contained a long-delay read.
This commit therefore inserts a trace_rcu_torture_read() event to flag
runs containing long delays.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The __call_rcu() comment about opportunistically noting grace period
beginnings and endings is obsolete. RCU still does such opportunistic
noting, but in __call_rcu_core() rather than __call_rcu(), and there
already is an appropriate comment in __call_rcu_core(). This commit
therefore removes the obsolete comment.
Reported-by: Michalis Kokologiannakis <mixaskok@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
In the deep past, rcu_check_callbacks() was only invoked if rcu_pending()
returned true. Which was fine, but these days rcu_check_callbacks()
is invoked unconditionally. This commit therefore removes the obsolete
sentence from the header comment.
Reported-by: Michalis Kokologiannakis <mixaskok@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
