€•      Œsphinx.addnodes”Œdocument”“”)”}”(Œ	rawsource”Œ ”Œchildren”]”(Œtranslations”ŒLanguagesNode”“”)”}”(hhh]”(h Œpending_xref”“”)”}”(hhh]”Œdocutils.nodes”ŒText”“”ŒChinese (Simplified)”…””}”Œparent”hsbaŒ
attributes”}”(Œids”]”Œclasses”]”Œnames”]”Œdupnames”]”Œbackrefs”]”Œ	refdomain”Œstd”Œreftype”Œdoc”Œ	reftarget”Œ!/translations/zh_CN/RCU/checklist”Œmodname”NŒ	classname”NŒrefexplicit”ˆuŒtagname”hhhubh)”}”(hhh]”hŒChinese (Traditional)”…””}”hh2sbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ!/translations/zh_TW/RCU/checklist”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubh)”}”(hhh]”hŒItalian”…””}”hhFsbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ!/translations/it_IT/RCU/checklist”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubh)”}”(hhh]”hŒJapanese”…””}”hhZsbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ!/translations/ja_JP/RCU/checklist”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubh)”}”(hhh]”hŒKorean”…””}”hhnsbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ!/translations/ko_KR/RCU/checklist”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubh)”}”(hhh]”hŒSpanish”…””}”hh‚sbah}”(h]”h ]”h"]”h$]”h&]”Œ	refdomain”h)Œreftype”h+Œ	reftarget”Œ!/translations/sp_SP/RCU/checklist”Œmodname”NŒ	classname”NŒrefexplicit”ˆuh1hhhubeh}”(h]”h ]”h"]”h$]”h&]”Œcurrent_language”ŒEnglish”uh1h
hhŒ	_document”hŒsource”NŒline”NubhŒcomment”“”)”}”(hŒ SPDX-License-Identifier: GPL-2.0”h]”hŒ SPDX-License-Identifier: GPL-2.0”…””}”hh£sbah}”(h]”h ]”h"]”h$]”h&]”Œ	xml:space”Œpreserve”uh1h¡hhhžhhŸŒ;/var/lib/git/docbuild/linux/Documentation/RCU/checklist.rst”h KubhŒsection”“”)”}”(hhh]”(hŒtitle”“”)”}”(hŒ Review Checklist for RCU Patches”h]”hŒ Review Checklist for RCU Patches”…””}”(hh»hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1h¹hh¶hžhhŸh³h KubhŒ	paragraph”“”)”}”(hXc  This document contains a checklist for producing and reviewing patches
that make use of RCU.  Violating any of the rules listed below will
result in the same sorts of problems that leaving out a locking primitive
would cause.  This list is based on experiences reviewing such patches
over a rather long period of time, but improvements are always welcome!”h]”hXc  This document contains a checklist for producing and reviewing patches
that make use of RCU.  Violating any of the rules listed below will
result in the same sorts of problems that leaving out a locking primitive
would cause.  This list is based on experiences reviewing such patches
over a rather long period of time, but improvements are always welcome!”…””}”(hhËhžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Khh¶hžhubhŒenumerated_list”“”)”}”(hhh]”(hŒ	list_item”“”)”}”(hX’  Is RCU being applied to a read-mostly situation?  If the data
structure is updated more than about 10% of the time, then you
should strongly consider some other approach, unless detailed
performance measurements show that RCU is nonetheless the right
tool for the job.  Yes, RCU does reduce read-side overhead by
increasing write-side overhead, which is exactly why normal uses
of RCU will do much more reading than updating.

Another exception is where performance is not an issue, and RCU
provides a simpler implementation.  An example of this situation
is the dynamic NMI code in the Linux 2.6 kernel, at least on
architectures where NMIs are rare.

Yet another exception is where the low real-time latency of RCU's
read-side primitives is critically important.

One final exception is where RCU readers are used to prevent
the ABA problem (https://en.wikipedia.org/wiki/ABA_problem)
for lockless updates.  This does result in the mildly
counter-intuitive situation where rcu_read_lock() and
rcu_read_unlock() are used to protect updates, however, this
approach can provide the same simplifications to certain types
of lockless algorithms that garbage collectors do.
”h]”(hÊ)”}”(hX©  Is RCU being applied to a read-mostly situation?  If the data
structure is updated more than about 10% of the time, then you
should strongly consider some other approach, unless detailed
performance measurements show that RCU is nonetheless the right
tool for the job.  Yes, RCU does reduce read-side overhead by
increasing write-side overhead, which is exactly why normal uses
of RCU will do much more reading than updating.”h]”hX©  Is RCU being applied to a read-mostly situation?  If the data
structure is updated more than about 10% of the time, then you
should strongly consider some other approach, unless detailed
performance measurements show that RCU is nonetheless the right
tool for the job.  Yes, RCU does reduce read-side overhead by
increasing write-side overhead, which is exactly why normal uses
of RCU will do much more reading than updating.”…””}”(hhähžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KhhàubhÊ)”}”(hŒàAnother exception is where performance is not an issue, and RCU
provides a simpler implementation.  An example of this situation
is the dynamic NMI code in the Linux 2.6 kernel, at least on
architectures where NMIs are rare.”h]”hŒàAnother exception is where performance is not an issue, and RCU
provides a simpler implementation.  An example of this situation
is the dynamic NMI code in the Linux 2.6 kernel, at least on
architectures where NMIs are rare.”…””}”(hhòhžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KhhàubhÊ)”}”(hŒoYet another exception is where the low real-time latency of RCU's
read-side primitives is critically important.”h]”hŒqYet another exception is where the low real-time latency of RCUâ€™s
read-side primitives is critically important.”…””}”(hj   hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KhhàubhÊ)”}”(hX“  One final exception is where RCU readers are used to prevent
the ABA problem (https://en.wikipedia.org/wiki/ABA_problem)
for lockless updates.  This does result in the mildly
counter-intuitive situation where rcu_read_lock() and
rcu_read_unlock() are used to protect updates, however, this
approach can provide the same simplifications to certain types
of lockless algorithms that garbage collectors do.”h]”(hŒNOne final exception is where RCU readers are used to prevent
the ABA problem (”…””}”(hj  hžhhŸNh NubhŒ	reference”“”)”}”(hŒ)https://en.wikipedia.org/wiki/ABA_problem”h]”hŒ)https://en.wikipedia.org/wiki/ABA_problem”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”Œrefuri”j  uh1j  hj  ubhX  )
for lockless updates.  This does result in the mildly
counter-intuitive situation where rcu_read_lock() and
rcu_read_unlock() are used to protect updates, however, this
approach can provide the same simplifications to certain types
of lockless algorithms that garbage collectors do.”…””}”(hj  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Khhàubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hX¯  Does the update code have proper mutual exclusion?

RCU does allow *readers* to run (almost) naked, but *writers* must
still use some sort of mutual exclusion, such as:

a.      locking,
b.      atomic operations, or
c.      restricting updates to a single task.

If you choose #b, be prepared to describe how you have handled
memory barriers on weakly ordered machines (pretty much all of
them -- even x86 allows later loads to be reordered to precede
earlier stores), and be prepared to explain why this added
complexity is worthwhile.  If you choose #c, be prepared to
explain how this single task does not become a major bottleneck
on large systems (for example, if the task is updating information
relating to itself that other tasks can read, there by definition
can be no bottleneck).  Note that the definition of "large" has
changed significantly:  Eight CPUs was "large" in the year 2000,
but a hundred CPUs was unremarkable in 2017.
”h]”(hÊ)”}”(hŒ2Does the update code have proper mutual exclusion?”h]”hŒ2Does the update code have proper mutual exclusion?”…””}”(hj;  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K&hj7  ubhÊ)”}”(hŒtRCU does allow *readers* to run (almost) naked, but *writers* must
still use some sort of mutual exclusion, such as:”h]”(hŒRCU does allow ”…””}”(hjI  hžhhŸNh NubhŒemphasis”“”)”}”(hŒ	*readers*”h]”hŒreaders”…””}”(hjS  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hjI  ubhŒ to run (almost) naked, but ”…””}”(hjI  hžhhŸNh NubjR  )”}”(hŒ	*writers*”h]”hŒwriters”…””}”(hje  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hjI  ubhŒ7 must
still use some sort of mutual exclusion, such as:”…””}”(hjI  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K(hj7  ubhÚ)”}”(hhh]”(hß)”}”(hŒlocking,”h]”hÊ)”}”(hj‚  h]”hŒlocking,”…””}”(hj„  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K+hj€  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj}  ubhß)”}”(hŒatomic operations, or”h]”hÊ)”}”(hj™  h]”hŒatomic operations, or”…””}”(hj›  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K,hj—  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj}  ubhß)”}”(hŒ&restricting updates to a single task.
”h]”hÊ)”}”(hŒ%restricting updates to a single task.”h]”hŒ%restricting updates to a single task.”…””}”(hj²  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K-hj®  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj}  ubeh}”(h]”h ]”h"]”h$]”h&]”Œenumtype”Œ
loweralpha”Œprefix”hŒsuffix”Œ.”uh1hÙhj7  ubhÊ)”}”(hX¦  If you choose #b, be prepared to describe how you have handled
memory barriers on weakly ordered machines (pretty much all of
them -- even x86 allows later loads to be reordered to precede
earlier stores), and be prepared to explain why this added
complexity is worthwhile.  If you choose #c, be prepared to
explain how this single task does not become a major bottleneck
on large systems (for example, if the task is updating information
relating to itself that other tasks can read, there by definition
can be no bottleneck).  Note that the definition of "large" has
changed significantly:  Eight CPUs was "large" in the year 2000,
but a hundred CPUs was unremarkable in 2017.”h]”hX®  If you choose #b, be prepared to describe how you have handled
memory barriers on weakly ordered machines (pretty much all of
them -- even x86 allows later loads to be reordered to precede
earlier stores), and be prepared to explain why this added
complexity is worthwhile.  If you choose #c, be prepared to
explain how this single task does not become a major bottleneck
on large systems (for example, if the task is updating information
relating to itself that other tasks can read, there by definition
can be no bottleneck).  Note that the definition of â€œlargeâ€ has
changed significantly:  Eight CPUs was â€œlargeâ€ in the year 2000,
but a hundred CPUs was unremarkable in 2017.”…””}”(hjÑ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K/hj7  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hXã  Do the RCU read-side critical sections make proper use of
rcu_read_lock() and friends?  These primitives are needed
to prevent grace periods from ending prematurely, which
could result in data being unceremoniously freed out from
under your read-side code, which can greatly increase the
actuarial risk of your kernel.

As a rough rule of thumb, any dereference of an RCU-protected
pointer must be covered by rcu_read_lock(), rcu_read_lock_bh(),
rcu_read_lock_sched(), or by the appropriate update-side lock.
Explicit disabling of preemption (preempt_disable(), for example)
can serve as rcu_read_lock_sched(), but is less readable and
prevents lockdep from detecting locking issues.  Acquiring a
spinlock also enters an RCU read-side critical section.

Please note that you *cannot* rely on code known to be built
only in non-preemptible kernels.  Such code can and will break,
especially in kernels built with CONFIG_PREEMPT_COUNT=y.

Letting RCU-protected pointers "leak" out of an RCU read-side
critical section is every bit as bad as letting them leak out
from under a lock.  Unless, of course, you have arranged some
other means of protection, such as a lock or a reference count
*before* letting them out of the RCU read-side critical section.
”h]”(hÊ)”}”(hX>  Do the RCU read-side critical sections make proper use of
rcu_read_lock() and friends?  These primitives are needed
to prevent grace periods from ending prematurely, which
could result in data being unceremoniously freed out from
under your read-side code, which can greatly increase the
actuarial risk of your kernel.”h]”hX>  Do the RCU read-side critical sections make proper use of
rcu_read_lock() and friends?  These primitives are needed
to prevent grace periods from ending prematurely, which
could result in data being unceremoniously freed out from
under your read-side code, which can greatly increase the
actuarial risk of your kernel.”…””}”(hjé  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K;hjå  ubhÊ)”}”(hX°  As a rough rule of thumb, any dereference of an RCU-protected
pointer must be covered by rcu_read_lock(), rcu_read_lock_bh(),
rcu_read_lock_sched(), or by the appropriate update-side lock.
Explicit disabling of preemption (preempt_disable(), for example)
can serve as rcu_read_lock_sched(), but is less readable and
prevents lockdep from detecting locking issues.  Acquiring a
spinlock also enters an RCU read-side critical section.”h]”hX°  As a rough rule of thumb, any dereference of an RCU-protected
pointer must be covered by rcu_read_lock(), rcu_read_lock_bh(),
rcu_read_lock_sched(), or by the appropriate update-side lock.
Explicit disabling of preemption (preempt_disable(), for example)
can serve as rcu_read_lock_sched(), but is less readable and
prevents lockdep from detecting locking issues.  Acquiring a
spinlock also enters an RCU read-side critical section.”…””}”(hj÷  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KBhjå  ubhÊ)”}”(hŒµPlease note that you *cannot* rely on code known to be built
only in non-preemptible kernels.  Such code can and will break,
especially in kernels built with CONFIG_PREEMPT_COUNT=y.”h]”(hŒPlease note that you ”…””}”(hj  hžhhŸNh NubjR  )”}”(hŒ*cannot*”h]”hŒcannot”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj  ubhŒ˜ rely on code known to be built
only in non-preemptible kernels.  Such code can and will break,
especially in kernels built with CONFIG_PREEMPT_COUNT=y.”…””}”(hj  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KJhjå  ubhÊ)”}”(hX9  Letting RCU-protected pointers "leak" out of an RCU read-side
critical section is every bit as bad as letting them leak out
from under a lock.  Unless, of course, you have arranged some
other means of protection, such as a lock or a reference count
*before* letting them out of the RCU read-side critical section.”h]”(hŒýLetting RCU-protected pointers â€œleakâ€ out of an RCU read-side
critical section is every bit as bad as letting them leak out
from under a lock.  Unless, of course, you have arranged some
other means of protection, such as a lock or a reference count
”…””}”(hj%  hžhhŸNh NubjR  )”}”(hŒ*before*”h]”hŒbefore”…””}”(hj-  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj%  ubhŒ8 letting them out of the RCU read-side critical section.”…””}”(hj%  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KNhjå  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hXN	  Does the update code tolerate concurrent accesses?

The whole point of RCU is to permit readers to run without
any locks or atomic operations.  This means that readers will
be running while updates are in progress.  There are a number
of ways to handle this concurrency, depending on the situation:

a.      Use the RCU variants of the list and hlist update
        primitives to add, remove, and replace elements on
        an RCU-protected list.  Alternatively, use the other
        RCU-protected data structures that have been added to
        the Linux kernel.

        This is almost always the best approach.

b.      Proceed as in (a) above, but also maintain per-element
        locks (that are acquired by both readers and writers)
        that guard per-element state.  Fields that the readers
        refrain from accessing can be guarded by some other lock
        acquired only by updaters, if desired.

        This also works quite well.

c.      Make updates appear atomic to readers.  For example,
        pointer updates to properly aligned fields will
        appear atomic, as will individual atomic primitives.
        Sequences of operations performed under a lock will *not*
        appear to be atomic to RCU readers, nor will sequences
        of multiple atomic primitives.  One alternative is to
        move multiple individual fields to a separate structure,
        thus solving the multiple-field problem by imposing an
        additional level of indirection.

        This can work, but is starting to get a bit tricky.

d.      Carefully order the updates and the reads so that readers
        see valid data at all phases of the update.  This is often
        more difficult than it sounds, especially given modern
        CPUs' tendency to reorder memory references.  One must
        usually liberally sprinkle memory-ordering operations
        through the code, making it difficult to understand and
        to test.  Where it works, it is better to use things
        like smp_store_release() and smp_load_acquire(), but in
        some cases the smp_mb() full memory barrier is required.

        As noted earlier, it is usually better to group the
        changing data into a separate structure, so that the
        change may be made to appear atomic by updating a pointer
        to reference a new structure containing updated values.
”h]”(hÊ)”}”(hŒ2Does the update code tolerate concurrent accesses?”h]”hŒ2Does the update code tolerate concurrent accesses?”…””}”(hjO  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KThjK  ubhÊ)”}”(hŒöThe whole point of RCU is to permit readers to run without
any locks or atomic operations.  This means that readers will
be running while updates are in progress.  There are a number
of ways to handle this concurrency, depending on the situation:”h]”hŒöThe whole point of RCU is to permit readers to run without
any locks or atomic operations.  This means that readers will
be running while updates are in progress.  There are a number
of ways to handle this concurrency, depending on the situation:”…””}”(hj]  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KVhjK  ubhÚ)”}”(hhh]”(hß)”}”(hX  Use the RCU variants of the list and hlist update
primitives to add, remove, and replace elements on
an RCU-protected list.  Alternatively, use the other
RCU-protected data structures that have been added to
the Linux kernel.

This is almost always the best approach.
”h]”(hÊ)”}”(hŒáUse the RCU variants of the list and hlist update
primitives to add, remove, and replace elements on
an RCU-protected list.  Alternatively, use the other
RCU-protected data structures that have been added to
the Linux kernel.”h]”hŒáUse the RCU variants of the list and hlist update
primitives to add, remove, and replace elements on
an RCU-protected list.  Alternatively, use the other
RCU-protected data structures that have been added to
the Linux kernel.”…””}”(hjr  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K[hjn  ubhÊ)”}”(hŒ(This is almost always the best approach.”h]”hŒ(This is almost always the best approach.”…””}”(hj€  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Kahjn  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjk  ubhß)”}”(hX!  Proceed as in (a) above, but also maintain per-element
locks (that are acquired by both readers and writers)
that guard per-element state.  Fields that the readers
refrain from accessing can be guarded by some other lock
acquired only by updaters, if desired.

This also works quite well.
”h]”(hÊ)”}”(hX  Proceed as in (a) above, but also maintain per-element
locks (that are acquired by both readers and writers)
that guard per-element state.  Fields that the readers
refrain from accessing can be guarded by some other lock
acquired only by updaters, if desired.”h]”hX  Proceed as in (a) above, but also maintain per-element
locks (that are acquired by both readers and writers)
that guard per-element state.  Fields that the readers
refrain from accessing can be guarded by some other lock
acquired only by updaters, if desired.”…””}”(hj˜  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Kchj”  ubhÊ)”}”(hŒThis also works quite well.”h]”hŒThis also works quite well.”…””}”(hj¦  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Kihj”  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjk  ubhß)”}”(hX  Make updates appear atomic to readers.  For example,
pointer updates to properly aligned fields will
appear atomic, as will individual atomic primitives.
Sequences of operations performed under a lock will *not*
appear to be atomic to RCU readers, nor will sequences
of multiple atomic primitives.  One alternative is to
move multiple individual fields to a separate structure,
thus solving the multiple-field problem by imposing an
additional level of indirection.

This can work, but is starting to get a bit tricky.
”h]”(hÊ)”}”(hXÑ  Make updates appear atomic to readers.  For example,
pointer updates to properly aligned fields will
appear atomic, as will individual atomic primitives.
Sequences of operations performed under a lock will *not*
appear to be atomic to RCU readers, nor will sequences
of multiple atomic primitives.  One alternative is to
move multiple individual fields to a separate structure,
thus solving the multiple-field problem by imposing an
additional level of indirection.”h]”(hŒÎMake updates appear atomic to readers.  For example,
pointer updates to properly aligned fields will
appear atomic, as will individual atomic primitives.
Sequences of operations performed under a lock will ”…””}”(hj¾  hžhhŸNh NubjR  )”}”(hŒ*not*”h]”hŒnot”…””}”(hjÆ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj¾  ubhŒþ
appear to be atomic to RCU readers, nor will sequences
of multiple atomic primitives.  One alternative is to
move multiple individual fields to a separate structure,
thus solving the multiple-field problem by imposing an
additional level of indirection.”…””}”(hj¾  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Kkhjº  ubhÊ)”}”(hŒ3This can work, but is starting to get a bit tricky.”h]”hŒ3This can work, but is starting to get a bit tricky.”…””}”(hjÞ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Kuhjº  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjk  ubhß)”}”(hXÓ  Carefully order the updates and the reads so that readers
see valid data at all phases of the update.  This is often
more difficult than it sounds, especially given modern
CPUs' tendency to reorder memory references.  One must
usually liberally sprinkle memory-ordering operations
through the code, making it difficult to understand and
to test.  Where it works, it is better to use things
like smp_store_release() and smp_load_acquire(), but in
some cases the smp_mb() full memory barrier is required.

As noted earlier, it is usually better to group the
changing data into a separate structure, so that the
change may be made to appear atomic by updating a pointer
to reference a new structure containing updated values.
”h]”(hÊ)”}”(hXö  Carefully order the updates and the reads so that readers
see valid data at all phases of the update.  This is often
more difficult than it sounds, especially given modern
CPUs' tendency to reorder memory references.  One must
usually liberally sprinkle memory-ordering operations
through the code, making it difficult to understand and
to test.  Where it works, it is better to use things
like smp_store_release() and smp_load_acquire(), but in
some cases the smp_mb() full memory barrier is required.”h]”hXø  Carefully order the updates and the reads so that readers
see valid data at all phases of the update.  This is often
more difficult than it sounds, especially given modern
CPUsâ€™ tendency to reorder memory references.  One must
usually liberally sprinkle memory-ordering operations
through the code, making it difficult to understand and
to test.  Where it works, it is better to use things
like smp_store_release() and smp_load_acquire(), but in
some cases the smp_mb() full memory barrier is required.”…””}”(hjö  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Kwhjò  ubhÊ)”}”(hŒÚAs noted earlier, it is usually better to group the
changing data into a separate structure, so that the
change may be made to appear atomic by updating a pointer
to reference a new structure containing updated values.”h]”hŒÚAs noted earlier, it is usually better to group the
changing data into a separate structure, so that the
change may be made to appear atomic by updating a pointer
to reference a new structure containing updated values.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Khjò  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjk  ubeh}”(h]”h ]”h"]”h$]”h&]”jÌ  jÍ  jÎ  hjÏ  jÐ  uh1hÙhjK  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸNh Nubhß)”}”(hXc  Weakly ordered CPUs pose special challenges.  Almost all CPUs
are weakly ordered -- even x86 CPUs allow later loads to be
reordered to precede earlier stores.  RCU code must take all of
the following measures to prevent memory-corruption problems:

a.      Readers must maintain proper ordering of their memory
        accesses.  The rcu_dereference() primitive ensures that
        the CPU picks up the pointer before it picks up the data
        that the pointer points to.  This really is necessary
        on Alpha CPUs.

        The rcu_dereference() primitive is also an excellent
        documentation aid, letting the person reading the
        code know exactly which pointers are protected by RCU.
        Please note that compilers can also reorder code, and
        they are becoming increasingly aggressive about doing
        just that.  The rcu_dereference() primitive therefore also
        prevents destructive compiler optimizations.  However,
        with a bit of devious creativity, it is possible to
        mishandle the return value from rcu_dereference().
        Please see rcu_dereference.rst for more information.

        The rcu_dereference() primitive is used by the
        various "_rcu()" list-traversal primitives, such
        as the list_for_each_entry_rcu().  Note that it is
        perfectly legal (if redundant) for update-side code to
        use rcu_dereference() and the "_rcu()" list-traversal
        primitives.  This is particularly useful in code that
        is common to readers and updaters.  However, lockdep
        will complain if you access rcu_dereference() outside
        of an RCU read-side critical section.  See lockdep.rst
        to learn what to do about this.

        Of course, neither rcu_dereference() nor the "_rcu()"
        list-traversal primitives can substitute for a good
        concurrency design coordinating among multiple updaters.

b.      If the list macros are being used, the list_add_tail_rcu()
        and list_add_rcu() primitives must be used in order
        to prevent weakly ordered machines from misordering
        structure initialization and pointer planting.
        Similarly, if the hlist macros are being used, the
        hlist_add_head_rcu() primitive is required.

c.      If the list macros are being used, the list_del_rcu()
        primitive must be used to keep list_del()'s pointer
        poisoning from inflicting toxic effects on concurrent
        readers.  Similarly, if the hlist macros are being used,
        the hlist_del_rcu() primitive is required.

        The list_replace_rcu() and hlist_replace_rcu() primitives
        may be used to replace an old structure with a new one
        in their respective types of RCU-protected lists.

d.      Rules similar to (4b) and (4c) apply to the "hlist_nulls"
        type of RCU-protected linked lists.

e.      Updates must ensure that initialization of a given
        structure happens before pointers to that structure are
        publicized.  Use the rcu_assign_pointer() primitive
        when publicizing a pointer to a structure that can
        be traversed by an RCU read-side critical section.
”h]”(hÊ)”}”(hŒ÷Weakly ordered CPUs pose special challenges.  Almost all CPUs
are weakly ordered -- even x86 CPUs allow later loads to be
reordered to precede earlier stores.  RCU code must take all of
the following measures to prevent memory-corruption problems:”h]”hŒ÷Weakly ordered CPUs pose special challenges.  Almost all CPUs
are weakly ordered -- even x86 CPUs allow later loads to be
reordered to precede earlier stores.  RCU code must take all of
the following measures to prevent memory-corruption problems:”…””}”(hj(  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K†hj$  ubhÚ)”}”(hhh]”(hß)”}”(hX¢  Readers must maintain proper ordering of their memory
accesses.  The rcu_dereference() primitive ensures that
the CPU picks up the pointer before it picks up the data
that the pointer points to.  This really is necessary
on Alpha CPUs.

The rcu_dereference() primitive is also an excellent
documentation aid, letting the person reading the
code know exactly which pointers are protected by RCU.
Please note that compilers can also reorder code, and
they are becoming increasingly aggressive about doing
just that.  The rcu_dereference() primitive therefore also
prevents destructive compiler optimizations.  However,
with a bit of devious creativity, it is possible to
mishandle the return value from rcu_dereference().
Please see rcu_dereference.rst for more information.

The rcu_dereference() primitive is used by the
various "_rcu()" list-traversal primitives, such
as the list_for_each_entry_rcu().  Note that it is
perfectly legal (if redundant) for update-side code to
use rcu_dereference() and the "_rcu()" list-traversal
primitives.  This is particularly useful in code that
is common to readers and updaters.  However, lockdep
will complain if you access rcu_dereference() outside
of an RCU read-side critical section.  See lockdep.rst
to learn what to do about this.

Of course, neither rcu_dereference() nor the "_rcu()"
list-traversal primitives can substitute for a good
concurrency design coordinating among multiple updaters.
”h]”(hÊ)”}”(hŒëReaders must maintain proper ordering of their memory
accesses.  The rcu_dereference() primitive ensures that
the CPU picks up the pointer before it picks up the data
that the pointer points to.  This really is necessary
on Alpha CPUs.”h]”hŒëReaders must maintain proper ordering of their memory
accesses.  The rcu_dereference() primitive ensures that
the CPU picks up the pointer before it picks up the data
that the pointer points to.  This really is necessary
on Alpha CPUs.”…””}”(hj=  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K‹hj9  ubhÊ)”}”(hX  The rcu_dereference() primitive is also an excellent
documentation aid, letting the person reading the
code know exactly which pointers are protected by RCU.
Please note that compilers can also reorder code, and
they are becoming increasingly aggressive about doing
just that.  The rcu_dereference() primitive therefore also
prevents destructive compiler optimizations.  However,
with a bit of devious creativity, it is possible to
mishandle the return value from rcu_dereference().
Please see rcu_dereference.rst for more information.”h]”hX  The rcu_dereference() primitive is also an excellent
documentation aid, letting the person reading the
code know exactly which pointers are protected by RCU.
Please note that compilers can also reorder code, and
they are becoming increasingly aggressive about doing
just that.  The rcu_dereference() primitive therefore also
prevents destructive compiler optimizations.  However,
with a bit of devious creativity, it is possible to
mishandle the return value from rcu_dereference().
Please see rcu_dereference.rst for more information.”…””}”(hjK  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K‘hj9  ubhÊ)”}”(hX÷  The rcu_dereference() primitive is used by the
various "_rcu()" list-traversal primitives, such
as the list_for_each_entry_rcu().  Note that it is
perfectly legal (if redundant) for update-side code to
use rcu_dereference() and the "_rcu()" list-traversal
primitives.  This is particularly useful in code that
is common to readers and updaters.  However, lockdep
will complain if you access rcu_dereference() outside
of an RCU read-side critical section.  See lockdep.rst
to learn what to do about this.”h]”hXÿ  The rcu_dereference() primitive is used by the
various â€œ_rcu()â€ list-traversal primitives, such
as the list_for_each_entry_rcu().  Note that it is
perfectly legal (if redundant) for update-side code to
use rcu_dereference() and the â€œ_rcu()â€ list-traversal
primitives.  This is particularly useful in code that
is common to readers and updaters.  However, lockdep
will complain if you access rcu_dereference() outside
of an RCU read-side critical section.  See lockdep.rst
to learn what to do about this.”…””}”(hjY  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Kœhj9  ubhÊ)”}”(hŒ¢Of course, neither rcu_dereference() nor the "_rcu()"
list-traversal primitives can substitute for a good
concurrency design coordinating among multiple updaters.”h]”hŒ¦Of course, neither rcu_dereference() nor the â€œ_rcu()â€
list-traversal primitives can substitute for a good
concurrency design coordinating among multiple updaters.”…””}”(hjg  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K§hj9  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj6  ubhß)”}”(hX1  If the list macros are being used, the list_add_tail_rcu()
and list_add_rcu() primitives must be used in order
to prevent weakly ordered machines from misordering
structure initialization and pointer planting.
Similarly, if the hlist macros are being used, the
hlist_add_head_rcu() primitive is required.
”h]”hÊ)”}”(hX0  If the list macros are being used, the list_add_tail_rcu()
and list_add_rcu() primitives must be used in order
to prevent weakly ordered machines from misordering
structure initialization and pointer planting.
Similarly, if the hlist macros are being used, the
hlist_add_head_rcu() primitive is required.”h]”hX0  If the list macros are being used, the list_add_tail_rcu()
and list_add_rcu() primitives must be used in order
to prevent weakly ordered machines from misordering
structure initialization and pointer planting.
Similarly, if the hlist macros are being used, the
hlist_add_head_rcu() primitive is required.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K«hj{  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj6  ubhß)”}”(hX¨  If the list macros are being used, the list_del_rcu()
primitive must be used to keep list_del()'s pointer
poisoning from inflicting toxic effects on concurrent
readers.  Similarly, if the hlist macros are being used,
the hlist_del_rcu() primitive is required.

The list_replace_rcu() and hlist_replace_rcu() primitives
may be used to replace an old structure with a new one
in their respective types of RCU-protected lists.
”h]”(hÊ)”}”(hX  If the list macros are being used, the list_del_rcu()
primitive must be used to keep list_del()'s pointer
poisoning from inflicting toxic effects on concurrent
readers.  Similarly, if the hlist macros are being used,
the hlist_del_rcu() primitive is required.”h]”hX  If the list macros are being used, the list_del_rcu()
primitive must be used to keep list_del()â€™s pointer
poisoning from inflicting toxic effects on concurrent
readers.  Similarly, if the hlist macros are being used,
the hlist_del_rcu() primitive is required.”…””}”(hj—  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K²hj“  ubhÊ)”}”(hŒ¢The list_replace_rcu() and hlist_replace_rcu() primitives
may be used to replace an old structure with a new one
in their respective types of RCU-protected lists.”h]”hŒ¢The list_replace_rcu() and hlist_replace_rcu() primitives
may be used to replace an old structure with a new one
in their respective types of RCU-protected lists.”…””}”(hj¥  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K¸hj“  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj6  ubhß)”}”(hŒ^Rules similar to (4b) and (4c) apply to the "hlist_nulls"
type of RCU-protected linked lists.
”h]”hÊ)”}”(hŒ]Rules similar to (4b) and (4c) apply to the "hlist_nulls"
type of RCU-protected linked lists.”h]”hŒaRules similar to (4b) and (4c) apply to the â€œhlist_nullsâ€
type of RCU-protected linked lists.”…””}”(hj½  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K¼hj¹  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj6  ubhß)”}”(hX  Updates must ensure that initialization of a given
structure happens before pointers to that structure are
publicized.  Use the rcu_assign_pointer() primitive
when publicizing a pointer to a structure that can
be traversed by an RCU read-side critical section.
”h]”hÊ)”}”(hX  Updates must ensure that initialization of a given
structure happens before pointers to that structure are
publicized.  Use the rcu_assign_pointer() primitive
when publicizing a pointer to a structure that can
be traversed by an RCU read-side critical section.”h]”hX  Updates must ensure that initialization of a given
structure happens before pointers to that structure are
publicized.  Use the rcu_assign_pointer() primitive
when publicizing a pointer to a structure that can
be traversed by an RCU read-side critical section.”…””}”(hjÕ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h K¿hjÑ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj6  ubeh}”(h]”h ]”h"]”h$]”h&]”jÌ  jÍ  jÎ  hjÏ  jÐ  uh1hÙhj$  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸNh Nubhß)”}”(hX¦  If any of call_rcu(), call_srcu(), call_rcu_tasks(), or
call_rcu_tasks_trace() is used, the callback function may be
invoked from softirq context, and in any case with bottom halves
disabled.  In particular, this callback function cannot block.
If you need the callback to block, run that code in a workqueue
handler scheduled from the callback.  The queue_rcu_work()
function does this for you in the case of call_rcu().
”h]”hÊ)”}”(hX¥  If any of call_rcu(), call_srcu(), call_rcu_tasks(), or
call_rcu_tasks_trace() is used, the callback function may be
invoked from softirq context, and in any case with bottom halves
disabled.  In particular, this callback function cannot block.
If you need the callback to block, run that code in a workqueue
handler scheduled from the callback.  The queue_rcu_work()
function does this for you in the case of call_rcu().”h]”hX¥  If any of call_rcu(), call_srcu(), call_rcu_tasks(), or
call_rcu_tasks_trace() is used, the callback function may be
invoked from softirq context, and in any case with bottom halves
disabled.  In particular, this callback function cannot block.
If you need the callback to block, run that code in a workqueue
handler scheduled from the callback.  The queue_rcu_work()
function does this for you in the case of call_rcu().”…””}”(hjù  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KÅhjõ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hX  Since synchronize_rcu() can block, it cannot be called
from any sort of irq context.  The same rule applies
for synchronize_srcu(), synchronize_rcu_expedited(),
synchronize_srcu_expedited(), synchronize_rcu_tasks(),
synchronize_rcu_tasks_rude(), and synchronize_rcu_tasks_trace().

The expedited forms of these primitives have the same semantics
as the non-expedited forms, but expediting is more CPU intensive.
Use of the expedited primitives should be restricted to rare
configuration-change operations that would not normally be
undertaken while a real-time workload is running.  Note that
IPI-sensitive real-time workloads can use the rcupdate.rcu_normal
kernel boot parameter to completely disable expedited grace
periods, though this might have performance implications.

In particular, if you find yourself invoking one of the expedited
primitives repeatedly in a loop, please do everyone a favor:
Restructure your code so that it batches the updates, allowing
a single non-expedited primitive to cover the entire batch.
This will very likely be faster than the loop containing the
expedited primitive, and will be much much easier on the rest
of the system, especially to real-time workloads running on the
rest of the system.  Alternatively, instead use asynchronous
primitives such as call_rcu().
”h]”(hÊ)”}”(hX  Since synchronize_rcu() can block, it cannot be called
from any sort of irq context.  The same rule applies
for synchronize_srcu(), synchronize_rcu_expedited(),
synchronize_srcu_expedited(), synchronize_rcu_tasks(),
synchronize_rcu_tasks_rude(), and synchronize_rcu_tasks_trace().”h]”hX  Since synchronize_rcu() can block, it cannot be called
from any sort of irq context.  The same rule applies
for synchronize_srcu(), synchronize_rcu_expedited(),
synchronize_srcu_expedited(), synchronize_rcu_tasks(),
synchronize_rcu_tasks_rude(), and synchronize_rcu_tasks_trace().”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KÍhj  ubhÊ)”}”(hXî  The expedited forms of these primitives have the same semantics
as the non-expedited forms, but expediting is more CPU intensive.
Use of the expedited primitives should be restricted to rare
configuration-change operations that would not normally be
undertaken while a real-time workload is running.  Note that
IPI-sensitive real-time workloads can use the rcupdate.rcu_normal
kernel boot parameter to completely disable expedited grace
periods, though this might have performance implications.”h]”hXî  The expedited forms of these primitives have the same semantics
as the non-expedited forms, but expediting is more CPU intensive.
Use of the expedited primitives should be restricted to rare
configuration-change operations that would not normally be
undertaken while a real-time workload is running.  Note that
IPI-sensitive real-time workloads can use the rcupdate.rcu_normal
kernel boot parameter to completely disable expedited grace
periods, though this might have performance implications.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KÓhj  ubhÊ)”}”(hX  In particular, if you find yourself invoking one of the expedited
primitives repeatedly in a loop, please do everyone a favor:
Restructure your code so that it batches the updates, allowing
a single non-expedited primitive to cover the entire batch.
This will very likely be faster than the loop containing the
expedited primitive, and will be much much easier on the rest
of the system, especially to real-time workloads running on the
rest of the system.  Alternatively, instead use asynchronous
primitives such as call_rcu().”h]”hX  In particular, if you find yourself invoking one of the expedited
primitives repeatedly in a loop, please do everyone a favor:
Restructure your code so that it batches the updates, allowing
a single non-expedited primitive to cover the entire batch.
This will very likely be faster than the loop containing the
expedited primitive, and will be much much easier on the rest
of the system, especially to real-time workloads running on the
rest of the system.  Alternatively, instead use asynchronous
primitives such as call_rcu().”…””}”(hj-  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KÜhj  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hXz  As of v4.20, a given kernel implements only one RCU flavor, which
is RCU-sched for PREEMPTION=n and RCU-preempt for PREEMPTION=y.
If the updater uses call_rcu() or synchronize_rcu(), then
the corresponding readers may use:  (1) rcu_read_lock() and
rcu_read_unlock(), (2) any pair of primitives that disables
and re-enables softirq, for example, rcu_read_lock_bh() and
rcu_read_unlock_bh(), or (3) any pair of primitives that disables
and re-enables preemption, for example, rcu_read_lock_sched() and
rcu_read_unlock_sched().  If the updater uses synchronize_srcu()
or call_srcu(), then the corresponding readers must use
srcu_read_lock() and srcu_read_unlock(), and with the same
srcu_struct.  The rules for the expedited RCU grace-period-wait
primitives are the same as for their non-expedited counterparts.

Similarly, it is necessary to correctly use the RCU Tasks flavors:

a.      If the updater uses synchronize_rcu_tasks() or
        call_rcu_tasks(), then the readers must refrain from
        executing voluntary context switches, that is, from
        blocking.

b.      If the updater uses call_rcu_tasks_trace()
        or synchronize_rcu_tasks_trace(), then the
        corresponding readers must use rcu_read_lock_trace()
        and rcu_read_unlock_trace().

c.      If an updater uses synchronize_rcu_tasks_rude(),
        then the corresponding readers must use anything that
        disables preemption, for example, preempt_disable()
        and preempt_enable().

Mixing things up will result in confusion and broken kernels, and
has even resulted in an exploitable security issue.  Therefore,
when using non-obvious pairs of primitives, commenting is
of course a must.  One example of non-obvious pairing is
the XDP feature in networking, which calls BPF programs from
network-driver NAPI (softirq) context.  BPF relies heavily on RCU
protection for its data structures, but because the BPF program
invocation happens entirely within a single local_bh_disable()
section in a NAPI poll cycle, this usage is safe.  The reason
that this usage is safe is that readers can use anything that
disables BH when updaters use call_rcu() or synchronize_rcu().
”h]”(hÊ)”}”(hX(  As of v4.20, a given kernel implements only one RCU flavor, which
is RCU-sched for PREEMPTION=n and RCU-preempt for PREEMPTION=y.
If the updater uses call_rcu() or synchronize_rcu(), then
the corresponding readers may use:  (1) rcu_read_lock() and
rcu_read_unlock(), (2) any pair of primitives that disables
and re-enables softirq, for example, rcu_read_lock_bh() and
rcu_read_unlock_bh(), or (3) any pair of primitives that disables
and re-enables preemption, for example, rcu_read_lock_sched() and
rcu_read_unlock_sched().  If the updater uses synchronize_srcu()
or call_srcu(), then the corresponding readers must use
srcu_read_lock() and srcu_read_unlock(), and with the same
srcu_struct.  The rules for the expedited RCU grace-period-wait
primitives are the same as for their non-expedited counterparts.”h]”hX(  As of v4.20, a given kernel implements only one RCU flavor, which
is RCU-sched for PREEMPTION=n and RCU-preempt for PREEMPTION=y.
If the updater uses call_rcu() or synchronize_rcu(), then
the corresponding readers may use:  (1) rcu_read_lock() and
rcu_read_unlock(), (2) any pair of primitives that disables
and re-enables softirq, for example, rcu_read_lock_bh() and
rcu_read_unlock_bh(), or (3) any pair of primitives that disables
and re-enables preemption, for example, rcu_read_lock_sched() and
rcu_read_unlock_sched().  If the updater uses synchronize_srcu()
or call_srcu(), then the corresponding readers must use
srcu_read_lock() and srcu_read_unlock(), and with the same
srcu_struct.  The rules for the expedited RCU grace-period-wait
primitives are the same as for their non-expedited counterparts.”…””}”(hjE  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KæhjA  ubhÊ)”}”(hŒBSimilarly, it is necessary to correctly use the RCU Tasks flavors:”h]”hŒBSimilarly, it is necessary to correctly use the RCU Tasks flavors:”…””}”(hjS  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h KôhjA  ubhÚ)”}”(hhh]”(hß)”}”(hŒ¢If the updater uses synchronize_rcu_tasks() or
call_rcu_tasks(), then the readers must refrain from
executing voluntary context switches, that is, from
blocking.
”h]”hÊ)”}”(hŒ¡If the updater uses synchronize_rcu_tasks() or
call_rcu_tasks(), then the readers must refrain from
executing voluntary context switches, that is, from
blocking.”h]”hŒ¡If the updater uses synchronize_rcu_tasks() or
call_rcu_tasks(), then the readers must refrain from
executing voluntary context switches, that is, from
blocking.”…””}”(hjh  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Köhjd  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhja  ubhß)”}”(hŒ¨If the updater uses call_rcu_tasks_trace()
or synchronize_rcu_tasks_trace(), then the
corresponding readers must use rcu_read_lock_trace()
and rcu_read_unlock_trace().
”h]”hÊ)”}”(hŒ§If the updater uses call_rcu_tasks_trace()
or synchronize_rcu_tasks_trace(), then the
corresponding readers must use rcu_read_lock_trace()
and rcu_read_unlock_trace().”h]”hŒ§If the updater uses call_rcu_tasks_trace()
or synchronize_rcu_tasks_trace(), then the
corresponding readers must use rcu_read_lock_trace()
and rcu_read_unlock_trace().”…””}”(hj€  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Kûhj|  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhja  ubhß)”}”(hŒ±If an updater uses synchronize_rcu_tasks_rude(),
then the corresponding readers must use anything that
disables preemption, for example, preempt_disable()
and preempt_enable().
”h]”hÊ)”}”(hŒ°If an updater uses synchronize_rcu_tasks_rude(),
then the corresponding readers must use anything that
disables preemption, for example, preempt_disable()
and preempt_enable().”h]”hŒ°If an updater uses synchronize_rcu_tasks_rude(),
then the corresponding readers must use anything that
disables preemption, for example, preempt_disable()
and preempt_enable().”…””}”(hj˜  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M hj”  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhja  ubeh}”(h]”h ]”h"]”h$]”h&]”jÌ  jÍ  jÎ  hjÏ  jÐ  uh1hÙhjA  ubhÊ)”}”(hX­  Mixing things up will result in confusion and broken kernels, and
has even resulted in an exploitable security issue.  Therefore,
when using non-obvious pairs of primitives, commenting is
of course a must.  One example of non-obvious pairing is
the XDP feature in networking, which calls BPF programs from
network-driver NAPI (softirq) context.  BPF relies heavily on RCU
protection for its data structures, but because the BPF program
invocation happens entirely within a single local_bh_disable()
section in a NAPI poll cycle, this usage is safe.  The reason
that this usage is safe is that readers can use anything that
disables BH when updaters use call_rcu() or synchronize_rcu().”h]”hX­  Mixing things up will result in confusion and broken kernels, and
has even resulted in an exploitable security issue.  Therefore,
when using non-obvious pairs of primitives, commenting is
of course a must.  One example of non-obvious pairing is
the XDP feature in networking, which calls BPF programs from
network-driver NAPI (softirq) context.  BPF relies heavily on RCU
protection for its data structures, but because the BPF program
invocation happens entirely within a single local_bh_disable()
section in a NAPI poll cycle, this usage is safe.  The reason
that this usage is safe is that readers can use anything that
disables BH when updaters use call_rcu() or synchronize_rcu().”…””}”(hj²  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MhjA  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hX  Although synchronize_rcu() is slower than is call_rcu(),
it usually results in simpler code.  So, unless update
performance is critically important, the updaters cannot block,
or the latency of synchronize_rcu() is visible from userspace,
synchronize_rcu() should be used in preference to call_rcu().
Furthermore, kfree_rcu() and kvfree_rcu() usually result
in even simpler code than does synchronize_rcu() without
synchronize_rcu()'s multi-millisecond latency.  So please take
advantage of kfree_rcu()'s and kvfree_rcu()'s "fire and forget"
memory-freeing capabilities where it applies.

An especially important property of the synchronize_rcu()
primitive is that it automatically self-limits: if grace periods
are delayed for whatever reason, then the synchronize_rcu()
primitive will correspondingly delay updates.  In contrast,
code using call_rcu() should explicitly limit update rate in
cases where grace periods are delayed, as failing to do so can
result in excessive realtime latencies or even OOM conditions.

Ways of gaining this self-limiting property when using call_rcu(),
kfree_rcu(), or kvfree_rcu() include:

a.      Keeping a count of the number of data-structure elements
        used by the RCU-protected data structure, including
        those waiting for a grace period to elapse.  Enforce a
        limit on this number, stalling updates as needed to allow
        previously deferred frees to complete.  Alternatively,
        limit only the number awaiting deferred free rather than
        the total number of elements.

        One way to stall the updates is to acquire the update-side
        mutex.  (Don't try this with a spinlock -- other CPUs
        spinning on the lock could prevent the grace period
        from ever ending.)  Another way to stall the updates
        is for the updates to use a wrapper function around
        the memory allocator, so that this wrapper function
        simulates OOM when there is too much memory awaiting an
        RCU grace period.  There are of course many other
        variations on this theme.

b.      Limiting update rate.  For example, if updates occur only
        once per hour, then no explicit rate limiting is
        required, unless your system is already badly broken.
        Older versions of the dcache subsystem take this approach,
        guarding updates with a global lock, limiting their rate.

c.      Trusted update -- if updates can only be done manually by
        superuser or some other trusted user, then it might not
        be necessary to automatically limit them.  The theory
        here is that superuser already has lots of ways to crash
        the machine.

d.      Periodically invoke rcu_barrier(), permitting a limited
        number of updates per grace period.

The same cautions apply to call_srcu(), call_rcu_tasks(), and
call_rcu_tasks_trace().  This is why there is an srcu_barrier(),
rcu_barrier_tasks(), and rcu_barrier_tasks_trace(), respectively.

Note that although these primitives do take action to avoid
memory exhaustion when any given CPU has too many callbacks,
a determined user or administrator can still exhaust memory.
This is especially the case if a system with a large number of
CPUs has been configured to offload all of its RCU callbacks onto
a single CPU, or if the system has relatively little free memory.
”h]”(hÊ)”}”(hXK  Although synchronize_rcu() is slower than is call_rcu(),
it usually results in simpler code.  So, unless update
performance is critically important, the updaters cannot block,
or the latency of synchronize_rcu() is visible from userspace,
synchronize_rcu() should be used in preference to call_rcu().
Furthermore, kfree_rcu() and kvfree_rcu() usually result
in even simpler code than does synchronize_rcu() without
synchronize_rcu()'s multi-millisecond latency.  So please take
advantage of kfree_rcu()'s and kvfree_rcu()'s "fire and forget"
memory-freeing capabilities where it applies.”h]”hXU  Although synchronize_rcu() is slower than is call_rcu(),
it usually results in simpler code.  So, unless update
performance is critically important, the updaters cannot block,
or the latency of synchronize_rcu() is visible from userspace,
synchronize_rcu() should be used in preference to call_rcu().
Furthermore, kfree_rcu() and kvfree_rcu() usually result
in even simpler code than does synchronize_rcu() without
synchronize_rcu()â€™s multi-millisecond latency.  So please take
advantage of kfree_rcu()â€™s and kvfree_rcu()â€™s â€œfire and forgetâ€
memory-freeing capabilities where it applies.”…””}”(hjÊ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MhjÆ  ubhÊ)”}”(hX­  An especially important property of the synchronize_rcu()
primitive is that it automatically self-limits: if grace periods
are delayed for whatever reason, then the synchronize_rcu()
primitive will correspondingly delay updates.  In contrast,
code using call_rcu() should explicitly limit update rate in
cases where grace periods are delayed, as failing to do so can
result in excessive realtime latencies or even OOM conditions.”h]”hX­  An especially important property of the synchronize_rcu()
primitive is that it automatically self-limits: if grace periods
are delayed for whatever reason, then the synchronize_rcu()
primitive will correspondingly delay updates.  In contrast,
code using call_rcu() should explicitly limit update rate in
cases where grace periods are delayed, as failing to do so can
result in excessive realtime latencies or even OOM conditions.”…””}”(hjØ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MhjÆ  ubhÊ)”}”(hŒhWays of gaining this self-limiting property when using call_rcu(),
kfree_rcu(), or kvfree_rcu() include:”h]”hŒhWays of gaining this self-limiting property when using call_rcu(),
kfree_rcu(), or kvfree_rcu() include:”…””}”(hjæ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M$hjÆ  ubhÚ)”}”(hhh]”(hß)”}”(hX3  Keeping a count of the number of data-structure elements
used by the RCU-protected data structure, including
those waiting for a grace period to elapse.  Enforce a
limit on this number, stalling updates as needed to allow
previously deferred frees to complete.  Alternatively,
limit only the number awaiting deferred free rather than
the total number of elements.

One way to stall the updates is to acquire the update-side
mutex.  (Don't try this with a spinlock -- other CPUs
spinning on the lock could prevent the grace period
from ever ending.)  Another way to stall the updates
is for the updates to use a wrapper function around
the memory allocator, so that this wrapper function
simulates OOM when there is too much memory awaiting an
RCU grace period.  There are of course many other
variations on this theme.
”h]”(hÊ)”}”(hXk  Keeping a count of the number of data-structure elements
used by the RCU-protected data structure, including
those waiting for a grace period to elapse.  Enforce a
limit on this number, stalling updates as needed to allow
previously deferred frees to complete.  Alternatively,
limit only the number awaiting deferred free rather than
the total number of elements.”h]”hXk  Keeping a count of the number of data-structure elements
used by the RCU-protected data structure, including
those waiting for a grace period to elapse.  Enforce a
limit on this number, stalling updates as needed to allow
previously deferred frees to complete.  Alternatively,
limit only the number awaiting deferred free rather than
the total number of elements.”…””}”(hjû  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M'hj÷  ubhÊ)”}”(hXÅ  One way to stall the updates is to acquire the update-side
mutex.  (Don't try this with a spinlock -- other CPUs
spinning on the lock could prevent the grace period
from ever ending.)  Another way to stall the updates
is for the updates to use a wrapper function around
the memory allocator, so that this wrapper function
simulates OOM when there is too much memory awaiting an
RCU grace period.  There are of course many other
variations on this theme.”h]”hXÇ  One way to stall the updates is to acquire the update-side
mutex.  (Donâ€™t try this with a spinlock -- other CPUs
spinning on the lock could prevent the grace period
from ever ending.)  Another way to stall the updates
is for the updates to use a wrapper function around
the memory allocator, so that this wrapper function
simulates OOM when there is too much memory awaiting an
RCU grace period.  There are of course many other
variations on this theme.”…””}”(hj	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M/hj÷  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjô  ubhß)”}”(hX  Limiting update rate.  For example, if updates occur only
once per hour, then no explicit rate limiting is
required, unless your system is already badly broken.
Older versions of the dcache subsystem take this approach,
guarding updates with a global lock, limiting their rate.
”h]”hÊ)”}”(hX  Limiting update rate.  For example, if updates occur only
once per hour, then no explicit rate limiting is
required, unless your system is already badly broken.
Older versions of the dcache subsystem take this approach,
guarding updates with a global lock, limiting their rate.”h]”hX  Limiting update rate.  For example, if updates occur only
once per hour, then no explicit rate limiting is
required, unless your system is already badly broken.
Older versions of the dcache subsystem take this approach,
guarding updates with a global lock, limiting their rate.”…””}”(hj!  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M9hj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjô  ubhß)”}”(hŒîTrusted update -- if updates can only be done manually by
superuser or some other trusted user, then it might not
be necessary to automatically limit them.  The theory
here is that superuser already has lots of ways to crash
the machine.
”h]”hÊ)”}”(hŒíTrusted update -- if updates can only be done manually by
superuser or some other trusted user, then it might not
be necessary to automatically limit them.  The theory
here is that superuser already has lots of ways to crash
the machine.”h]”hŒíTrusted update -- if updates can only be done manually by
superuser or some other trusted user, then it might not
be necessary to automatically limit them.  The theory
here is that superuser already has lots of ways to crash
the machine.”…””}”(hj9  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M?hj5  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjô  ubhß)”}”(hŒ\Periodically invoke rcu_barrier(), permitting a limited
number of updates per grace period.
”h]”hÊ)”}”(hŒ[Periodically invoke rcu_barrier(), permitting a limited
number of updates per grace period.”h]”hŒ[Periodically invoke rcu_barrier(), permitting a limited
number of updates per grace period.”…””}”(hjQ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MEhjM  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjô  ubeh}”(h]”h ]”h"]”h$]”h&]”jÌ  jÍ  jÎ  hjÏ  jÐ  uh1hÙhjÆ  ubhÊ)”}”(hŒÀThe same cautions apply to call_srcu(), call_rcu_tasks(), and
call_rcu_tasks_trace().  This is why there is an srcu_barrier(),
rcu_barrier_tasks(), and rcu_barrier_tasks_trace(), respectively.”h]”hŒÀThe same cautions apply to call_srcu(), call_rcu_tasks(), and
call_rcu_tasks_trace().  This is why there is an srcu_barrier(),
rcu_barrier_tasks(), and rcu_barrier_tasks_trace(), respectively.”…””}”(hjk  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MHhjÆ  ubhÊ)”}”(hXx  Note that although these primitives do take action to avoid
memory exhaustion when any given CPU has too many callbacks,
a determined user or administrator can still exhaust memory.
This is especially the case if a system with a large number of
CPUs has been configured to offload all of its RCU callbacks onto
a single CPU, or if the system has relatively little free memory.”h]”hXx  Note that although these primitives do take action to avoid
memory exhaustion when any given CPU has too many callbacks,
a determined user or administrator can still exhaust memory.
This is especially the case if a system with a large number of
CPUs has been configured to offload all of its RCU callbacks onto
a single CPU, or if the system has relatively little free memory.”…””}”(hjy  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MLhjÆ  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hXŸ  All RCU list-traversal primitives, which include
rcu_dereference(), list_for_each_entry_rcu(), and
list_for_each_safe_rcu(), must be either within an RCU read-side
critical section or must be protected by appropriate update-side
locks.  RCU read-side critical sections are delimited by
rcu_read_lock() and rcu_read_unlock(), or by similar primitives
such as rcu_read_lock_bh() and rcu_read_unlock_bh(), in which
case the matching rcu_dereference() primitive must be used in
order to keep lockdep happy, in this case, rcu_dereference_bh().

The reason that it is permissible to use RCU list-traversal
primitives when the update-side lock is held is that doing so
can be quite helpful in reducing code bloat when common code is
shared between readers and updaters.  Additional primitives
are provided for this case, as discussed in lockdep.rst.

One exception to this rule is when data is only ever added to
the linked data structure, and is never removed during any
time that readers might be accessing that structure.  In such
cases, READ_ONCE() may be used in place of rcu_dereference()
and the read-side markers (rcu_read_lock() and rcu_read_unlock(),
for example) may be omitted.
”h]”(hÊ)”}”(hX  All RCU list-traversal primitives, which include
rcu_dereference(), list_for_each_entry_rcu(), and
list_for_each_safe_rcu(), must be either within an RCU read-side
critical section or must be protected by appropriate update-side
locks.  RCU read-side critical sections are delimited by
rcu_read_lock() and rcu_read_unlock(), or by similar primitives
such as rcu_read_lock_bh() and rcu_read_unlock_bh(), in which
case the matching rcu_dereference() primitive must be used in
order to keep lockdep happy, in this case, rcu_dereference_bh().”•‡–      h]”hX  All RCU list-traversal primitives, which include
rcu_dereference(), list_for_each_entry_rcu(), and
list_for_each_safe_rcu(), must be either within an RCU read-side
critical section or must be protected by appropriate update-side
locks.  RCU read-side critical sections are delimited by
rcu_read_lock() and rcu_read_unlock(), or by similar primitives
such as rcu_read_lock_bh() and rcu_read_unlock_bh(), in which
case the matching rcu_dereference() primitive must be used in
order to keep lockdep happy, in this case, rcu_dereference_bh().”…””}”(hj‘  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MShj  ubhÊ)”}”(hX.  The reason that it is permissible to use RCU list-traversal
primitives when the update-side lock is held is that doing so
can be quite helpful in reducing code bloat when common code is
shared between readers and updaters.  Additional primitives
are provided for this case, as discussed in lockdep.rst.”h]”hX.  The reason that it is permissible to use RCU list-traversal
primitives when the update-side lock is held is that doing so
can be quite helpful in reducing code bloat when common code is
shared between readers and updaters.  Additional primitives
are provided for this case, as discussed in lockdep.rst.”…””}”(hjŸ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M]hj  ubhÊ)”}”(hXR  One exception to this rule is when data is only ever added to
the linked data structure, and is never removed during any
time that readers might be accessing that structure.  In such
cases, READ_ONCE() may be used in place of rcu_dereference()
and the read-side markers (rcu_read_lock() and rcu_read_unlock(),
for example) may be omitted.”h]”hXR  One exception to this rule is when data is only ever added to
the linked data structure, and is never removed during any
time that readers might be accessing that structure.  In such
cases, READ_ONCE() may be used in place of rcu_dereference()
and the read-side markers (rcu_read_lock() and rcu_read_unlock(),
for example) may be omitted.”…””}”(hj­  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mchj  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hX3  Conversely, if you are in an RCU read-side critical section,
and you don't hold the appropriate update-side lock, you *must*
use the "_rcu()" variants of the list macros.  Failing to do so
will break Alpha, cause aggressive compilers to generate bad code,
and confuse people trying to understand your code.
”h]”hÊ)”}”(hX2  Conversely, if you are in an RCU read-side critical section,
and you don't hold the appropriate update-side lock, you *must*
use the "_rcu()" variants of the list macros.  Failing to do so
will break Alpha, cause aggressive compilers to generate bad code,
and confuse people trying to understand your code.”h]”(hŒxConversely, if you are in an RCU read-side critical section,
and you donâ€™t hold the appropriate update-side lock, you ”…””}”(hjÅ  hžhhŸNh NubjR  )”}”(hŒ*must*”h]”hŒmust”…””}”(hjÍ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hjÅ  ubhŒº
use the â€œ_rcu()â€ variants of the list macros.  Failing to do so
will break Alpha, cause aggressive compilers to generate bad code,
and confuse people trying to understand your code.”…””}”(hjÅ  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MjhjÁ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hXE  Any lock acquired by an RCU callback must be acquired elsewhere
with softirq disabled, e.g., via spin_lock_bh().  Failing to
disable softirq on a given acquisition of that lock will result
in deadlock as soon as the RCU softirq handler happens to run
your RCU callback while interrupting that acquisition's critical
section.
”h]”hÊ)”}”(hXD  Any lock acquired by an RCU callback must be acquired elsewhere
with softirq disabled, e.g., via spin_lock_bh().  Failing to
disable softirq on a given acquisition of that lock will result
in deadlock as soon as the RCU softirq handler happens to run
your RCU callback while interrupting that acquisition's critical
section.”h]”hXF  Any lock acquired by an RCU callback must be acquired elsewhere
with softirq disabled, e.g., via spin_lock_bh().  Failing to
disable softirq on a given acquisition of that lock will result
in deadlock as soon as the RCU softirq handler happens to run
your RCU callback while interrupting that acquisitionâ€™s critical
section.”…””}”(hjï  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mphjë  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hXv  RCU callbacks can be and are executed in parallel.  In many cases,
the callback code simply wrappers around kfree(), so that this
is not an issue (or, more accurately, to the extent that it is
an issue, the memory-allocator locking handles it).  However,
if the callbacks do manipulate a shared data structure, they
must use whatever locking or other synchronization is required
to safely access and/or modify that data structure.

Do not assume that RCU callbacks will be executed on the same
CPU that executed the corresponding call_rcu(), call_srcu(),
call_rcu_tasks(), or call_rcu_tasks_trace().  For example, if
a given CPU goes offline while having an RCU callback pending,
then that RCU callback will execute on some surviving CPU.
(If this was not the case, a self-spawning RCU callback would
prevent the victim CPU from ever going offline.)  Furthermore,
CPUs designated by rcu_nocbs= might well *always* have their
RCU callbacks executed on some other CPUs, in fact, for some
real-time workloads, this is the whole point of using the
rcu_nocbs= kernel boot parameter.

In addition, do not assume that callbacks queued in a given order
will be invoked in that order, even if they all are queued on the
same CPU.  Furthermore, do not assume that same-CPU callbacks will
be invoked serially.  For example, in recent kernels, CPUs can be
switched between offloaded and de-offloaded callback invocation,
and while a given CPU is undergoing such a switch, its callbacks
might be concurrently invoked by that CPU's softirq handler and
that CPU's rcuo kthread.  At such times, that CPU's callbacks
might be executed both concurrently and out of order.
”h]”(hÊ)”}”(hX®  RCU callbacks can be and are executed in parallel.  In many cases,
the callback code simply wrappers around kfree(), so that this
is not an issue (or, more accurately, to the extent that it is
an issue, the memory-allocator locking handles it).  However,
if the callbacks do manipulate a shared data structure, they
must use whatever locking or other synchronization is required
to safely access and/or modify that data structure.”h]”hX®  RCU callbacks can be and are executed in parallel.  In many cases,
the callback code simply wrappers around kfree(), so that this
is not an issue (or, more accurately, to the extent that it is
an issue, the memory-allocator locking handles it).  However,
if the callbacks do manipulate a shared data structure, they
must use whatever locking or other synchronization is required
to safely access and/or modify that data structure.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mwhj  ubhÊ)”}”(hX…  Do not assume that RCU callbacks will be executed on the same
CPU that executed the corresponding call_rcu(), call_srcu(),
call_rcu_tasks(), or call_rcu_tasks_trace().  For example, if
a given CPU goes offline while having an RCU callback pending,
then that RCU callback will execute on some surviving CPU.
(If this was not the case, a self-spawning RCU callback would
prevent the victim CPU from ever going offline.)  Furthermore,
CPUs designated by rcu_nocbs= might well *always* have their
RCU callbacks executed on some other CPUs, in fact, for some
real-time workloads, this is the whole point of using the
rcu_nocbs= kernel boot parameter.”h]”(hXÙ  Do not assume that RCU callbacks will be executed on the same
CPU that executed the corresponding call_rcu(), call_srcu(),
call_rcu_tasks(), or call_rcu_tasks_trace().  For example, if
a given CPU goes offline while having an RCU callback pending,
then that RCU callback will execute on some surviving CPU.
(If this was not the case, a self-spawning RCU callback would
prevent the victim CPU from ever going offline.)  Furthermore,
CPUs designated by rcu_nocbs= might well ”…””}”(hj  hžhhŸNh NubjR  )”}”(hŒ*always*”h]”hŒalways”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj  ubhŒ¤ have their
RCU callbacks executed on some other CPUs, in fact, for some
real-time workloads, this is the whole point of using the
rcu_nocbs= kernel boot parameter.”…””}”(hj  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj  ubhÊ)”}”(hX>  In addition, do not assume that callbacks queued in a given order
will be invoked in that order, even if they all are queued on the
same CPU.  Furthermore, do not assume that same-CPU callbacks will
be invoked serially.  For example, in recent kernels, CPUs can be
switched between offloaded and de-offloaded callback invocation,
and while a given CPU is undergoing such a switch, its callbacks
might be concurrently invoked by that CPU's softirq handler and
that CPU's rcuo kthread.  At such times, that CPU's callbacks
might be executed both concurrently and out of order.”h]”hXD  In addition, do not assume that callbacks queued in a given order
will be invoked in that order, even if they all are queued on the
same CPU.  Furthermore, do not assume that same-CPU callbacks will
be invoked serially.  For example, in recent kernels, CPUs can be
switched between offloaded and de-offloaded callback invocation,
and while a given CPU is undergoing such a switch, its callbacks
might be concurrently invoked by that CPUâ€™s softirq handler and
that CPUâ€™s rcuo kthread.  At such times, that CPUâ€™s callbacks
might be executed both concurrently and out of order.”…””}”(hj5  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M‹hj  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hX>
  Unlike most flavors of RCU, it *is* permissible to block in an
SRCU read-side critical section (demarked by srcu_read_lock()
and srcu_read_unlock()), hence the "SRCU": "sleepable RCU".
Please note that if you don't need to sleep in read-side critical
sections, you should be using RCU rather than SRCU, because RCU
is almost always faster and easier to use than is SRCU.

Also unlike other forms of RCU, explicit initialization and
cleanup is required either at build time via DEFINE_SRCU()
or DEFINE_STATIC_SRCU() or at runtime via init_srcu_struct()
and cleanup_srcu_struct().  These last two are passed a
"struct srcu_struct" that defines the scope of a given
SRCU domain.  Once initialized, the srcu_struct is passed
to srcu_read_lock(), srcu_read_unlock() synchronize_srcu(),
synchronize_srcu_expedited(), and call_srcu().  A given
synchronize_srcu() waits only for SRCU read-side critical
sections governed by srcu_read_lock() and srcu_read_unlock()
calls that have been passed the same srcu_struct.  This property
is what makes sleeping read-side critical sections tolerable --
a given subsystem delays only its own updates, not those of other
subsystems using SRCU.  Therefore, SRCU is less prone to OOM the
system than RCU would be if RCU's read-side critical sections
were permitted to sleep.

The ability to sleep in read-side critical sections does not
come for free.  First, corresponding srcu_read_lock() and
srcu_read_unlock() calls must be passed the same srcu_struct.
Second, grace-period-detection overhead is amortized only
over those updates sharing a given srcu_struct, rather than
being globally amortized as they are for other forms of RCU.
Therefore, SRCU should be used in preference to rw_semaphore
only in extremely read-intensive situations, or in situations
requiring SRCU's read-side deadlock immunity or low read-side
realtime latency.  You should also consider percpu_rw_semaphore
when you need lightweight readers.

SRCU's expedited primitive (synchronize_srcu_expedited())
never sends IPIs to other CPUs, so it is easier on
real-time workloads than is synchronize_rcu_expedited().

It is also permissible to sleep in RCU Tasks Trace read-side
critical section, which are delimited by rcu_read_lock_trace() and
rcu_read_unlock_trace().  However, this is a specialized flavor
of RCU, and you should not use it without first checking with
its current users.  In most cases, you should instead use SRCU.

Note that rcu_assign_pointer() relates to SRCU just as it does to
other forms of RCU, but instead of rcu_dereference() you should
use srcu_dereference() in order to avoid lockdep splats.
”h]”(hÊ)”}”(hXr  Unlike most flavors of RCU, it *is* permissible to block in an
SRCU read-side critical section (demarked by srcu_read_lock()
and srcu_read_unlock()), hence the "SRCU": "sleepable RCU".
Please note that if you don't need to sleep in read-side critical
sections, you should be using RCU rather than SRCU, because RCU
is almost always faster and easier to use than is SRCU.”h]”(hŒUnlike most flavors of RCU, it ”…””}”(hjM  hžhhŸNh NubjR  )”}”(hŒ*is*”h]”hŒis”…””}”(hjU  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hjM  ubhXY   permissible to block in an
SRCU read-side critical section (demarked by srcu_read_lock()
and srcu_read_unlock()), hence the â€œSRCUâ€: â€œsleepable RCUâ€.
Please note that if you donâ€™t need to sleep in read-side critical
sections, you should be using RCU rather than SRCU, because RCU
is almost always faster and easier to use than is SRCU.”…””}”(hjM  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M•hjI  ubhÊ)”}”(hX¢  Also unlike other forms of RCU, explicit initialization and
cleanup is required either at build time via DEFINE_SRCU()
or DEFINE_STATIC_SRCU() or at runtime via init_srcu_struct()
and cleanup_srcu_struct().  These last two are passed a
"struct srcu_struct" that defines the scope of a given
SRCU domain.  Once initialized, the srcu_struct is passed
to srcu_read_lock(), srcu_read_unlock() synchronize_srcu(),
synchronize_srcu_expedited(), and call_srcu().  A given
synchronize_srcu() waits only for SRCU read-side critical
sections governed by srcu_read_lock() and srcu_read_unlock()
calls that have been passed the same srcu_struct.  This property
is what makes sleeping read-side critical sections tolerable --
a given subsystem delays only its own updates, not those of other
subsystems using SRCU.  Therefore, SRCU is less prone to OOM the
system than RCU would be if RCU's read-side critical sections
were permitted to sleep.”h]”hX¨  Also unlike other forms of RCU, explicit initialization and
cleanup is required either at build time via DEFINE_SRCU()
or DEFINE_STATIC_SRCU() or at runtime via init_srcu_struct()
and cleanup_srcu_struct().  These last two are passed a
â€œstruct srcu_structâ€ that defines the scope of a given
SRCU domain.  Once initialized, the srcu_struct is passed
to srcu_read_lock(), srcu_read_unlock() synchronize_srcu(),
synchronize_srcu_expedited(), and call_srcu().  A given
synchronize_srcu() waits only for SRCU read-side critical
sections governed by srcu_read_lock() and srcu_read_unlock()
calls that have been passed the same srcu_struct.  This property
is what makes sleeping read-side critical sections tolerable --
a given subsystem delays only its own updates, not those of other
subsystems using SRCU.  Therefore, SRCU is less prone to OOM the
system than RCU would be if RCUâ€™s read-side critical sections
were permitted to sleep.”…””}”(hjm  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MœhjI  ubhÊ)”}”(hXƒ  The ability to sleep in read-side critical sections does not
come for free.  First, corresponding srcu_read_lock() and
srcu_read_unlock() calls must be passed the same srcu_struct.
Second, grace-period-detection overhead is amortized only
over those updates sharing a given srcu_struct, rather than
being globally amortized as they are for other forms of RCU.
Therefore, SRCU should be used in preference to rw_semaphore
only in extremely read-intensive situations, or in situations
requiring SRCU's read-side deadlock immunity or low read-side
realtime latency.  You should also consider percpu_rw_semaphore
when you need lightweight readers.”h]”hX…  The ability to sleep in read-side critical sections does not
come for free.  First, corresponding srcu_read_lock() and
srcu_read_unlock() calls must be passed the same srcu_struct.
Second, grace-period-detection overhead is amortized only
over those updates sharing a given srcu_struct, rather than
being globally amortized as they are for other forms of RCU.
Therefore, SRCU should be used in preference to rw_semaphore
only in extremely read-intensive situations, or in situations
requiring SRCUâ€™s read-side deadlock immunity or low read-side
realtime latency.  You should also consider percpu_rw_semaphore
when you need lightweight readers.”…””}”(hj{  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M­hjI  ubhÊ)”}”(hŒ¥SRCU's expedited primitive (synchronize_srcu_expedited())
never sends IPIs to other CPUs, so it is easier on
real-time workloads than is synchronize_rcu_expedited().”h]”hŒ§SRCUâ€™s expedited primitive (synchronize_srcu_expedited())
never sends IPIs to other CPUs, so it is easier on
real-time workloads than is synchronize_rcu_expedited().”…””}”(hj‰  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M¹hjI  ubhÊ)”}”(hX=  It is also permissible to sleep in RCU Tasks Trace read-side
critical section, which are delimited by rcu_read_lock_trace() and
rcu_read_unlock_trace().  However, this is a specialized flavor
of RCU, and you should not use it without first checking with
its current users.  In most cases, you should instead use SRCU.”h]”hX=  It is also permissible to sleep in RCU Tasks Trace read-side
critical section, which are delimited by rcu_read_lock_trace() and
rcu_read_unlock_trace().  However, this is a specialized flavor
of RCU, and you should not use it without first checking with
its current users.  In most cases, you should instead use SRCU.”…””}”(hj—  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M½hjI  ubhÊ)”}”(hŒºNote that rcu_assign_pointer() relates to SRCU just as it does to
other forms of RCU, but instead of rcu_dereference() you should
use srcu_dereference() in order to avoid lockdep splats.”h]”hŒºNote that rcu_assign_pointer() relates to SRCU just as it does to
other forms of RCU, but instead of rcu_dereference() you should
use srcu_dereference() in order to avoid lockdep splats.”…””}”(hj¥  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MÃhjI  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hX  The whole point of call_rcu(), synchronize_rcu(), and friends
is to wait until all pre-existing readers have finished before
carrying out some otherwise-destructive operation.  It is
therefore critically important to *first* remove any path
that readers can follow that could be affected by the
destructive operation, and *only then* invoke call_rcu(),
synchronize_rcu(), or friends.

Because these primitives only wait for pre-existing readers, it
is the caller's responsibility to guarantee that any subsequent
readers will execute safely.
”h]”(hÊ)”}”(hX  The whole point of call_rcu(), synchronize_rcu(), and friends
is to wait until all pre-existing readers have finished before
carrying out some otherwise-destructive operation.  It is
therefore critically important to *first* remove any path
that readers can follow that could be affected by the
destructive operation, and *only then* invoke call_rcu(),
synchronize_rcu(), or friends.”h]”(hŒÙThe whole point of call_rcu(), synchronize_rcu(), and friends
is to wait until all pre-existing readers have finished before
carrying out some otherwise-destructive operation.  It is
therefore critically important to ”…””}”(hj½  hžhhŸNh NubjR  )”}”(hŒ*first*”h]”hŒfirst”…””}”(hjÅ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj½  ubhŒb remove any path
that readers can follow that could be affected by the
destructive operation, and ”…””}”(hj½  hžhhŸNh NubjR  )”}”(hŒ*only then*”h]”hŒ	only then”…””}”(hj×  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj½  ubhŒ2 invoke call_rcu(),
synchronize_rcu(), or friends.”…””}”(hj½  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MÇhj¹  ubhÊ)”}”(hŒœBecause these primitives only wait for pre-existing readers, it
is the caller's responsibility to guarantee that any subsequent
readers will execute safely.”h]”hŒžBecause these primitives only wait for pre-existing readers, it
is the callerâ€™s responsibility to guarantee that any subsequent
readers will execute safely.”…””}”(hjï  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MÏhj¹  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hX¡  The various RCU read-side primitives do *not* necessarily contain
memory barriers.  You should therefore plan for the CPU
and the compiler to freely reorder code into and out of RCU
read-side critical sections.  It is the responsibility of the
RCU update-side primitives to deal with this.

For SRCU readers, you can use smp_mb__after_srcu_read_unlock()
immediately after an srcu_read_unlock() to get a full barrier.
”h]”(hÊ)”}”(hX!  The various RCU read-side primitives do *not* necessarily contain
memory barriers.  You should therefore plan for the CPU
and the compiler to freely reorder code into and out of RCU
read-side critical sections.  It is the responsibility of the
RCU update-side primitives to deal with this.”h]”(hŒ(The various RCU read-side primitives do ”…””}”(hj  hžhhŸNh NubjR  )”}”(hŒ*not*”h]”hŒnot”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj  ubhŒô necessarily contain
memory barriers.  You should therefore plan for the CPU
and the compiler to freely reorder code into and out of RCU
read-side critical sections.  It is the responsibility of the
RCU update-side primitives to deal with this.”…””}”(hj  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MÓhj  ubhÊ)”}”(hŒ}For SRCU readers, you can use smp_mb__after_srcu_read_unlock()
immediately after an srcu_read_unlock() to get a full barrier.”h]”hŒ}For SRCU readers, you can use smp_mb__after_srcu_read_unlock()
immediately after an srcu_read_unlock() to get a full barrier.”…””}”(hj'  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MÙhj  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hXð  Use CONFIG_PROVE_LOCKING, CONFIG_DEBUG_OBJECTS_RCU_HEAD, and the
__rcu sparse checks to validate your RCU code.  These can help
find problems as follows:

CONFIG_PROVE_LOCKING:
        check that accesses to RCU-protected data structures
        are carried out under the proper RCU read-side critical
        section, while holding the right combination of locks,
        or whatever other conditions are appropriate.

CONFIG_DEBUG_OBJECTS_RCU_HEAD:
        check that you don't pass the same object to call_rcu()
        (or friends) before an RCU grace period has elapsed
        since the last time that you passed that same object to
        call_rcu() (or friends).

CONFIG_RCU_STRICT_GRACE_PERIOD:
        combine with KASAN to check for pointers leaked out
        of RCU read-side critical sections.  This Kconfig
        option is tough on both performance and scalability,
        and so is limited to four-CPU systems.

__rcu sparse checks:
        tag the pointer to the RCU-protected data structure
        with __rcu, and sparse will warn you if you access that
        pointer without the services of one of the variants
        of rcu_dereference().

These debugging aids can help you find problems that are
otherwise extremely difficult to spot.
”h]”(hÊ)”}”(hŒ™Use CONFIG_PROVE_LOCKING, CONFIG_DEBUG_OBJECTS_RCU_HEAD, and the
__rcu sparse checks to validate your RCU code.  These can help
find problems as follows:”h]”hŒ™Use CONFIG_PROVE_LOCKING, CONFIG_DEBUG_OBJECTS_RCU_HEAD, and the
__rcu sparse checks to validate your RCU code.  These can help
find problems as follows:”…””}”(hj?  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MÜhj;  ubhŒdefinition_list”“”)”}”(hhh]”(hŒdefinition_list_item”“”)”}”(hŒèCONFIG_PROVE_LOCKING:
check that accesses to RCU-protected data structures
are carried out under the proper RCU read-side critical
section, while holding the right combination of locks,
or whatever other conditions are appropriate.
”h]”(hŒterm”“”)”}”(hŒCONFIG_PROVE_LOCKING:”h]”hŒCONFIG_PROVE_LOCKING:”…””}”(hjZ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jX  hŸh³h MähjT  ubhŒ
definition”“”)”}”(hhh]”hÊ)”}”(hŒÑcheck that accesses to RCU-protected data structures
are carried out under the proper RCU read-side critical
section, while holding the right combination of locks,
or whatever other conditions are appropriate.”h]”hŒÑcheck that accesses to RCU-protected data structures
are carried out under the proper RCU read-side critical
section, while holding the right combination of locks,
or whatever other conditions are appropriate.”…””}”(hjm  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Máhjj  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jh  hjT  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jR  hŸh³h MähjO  ubjS  )”}”(hŒÜCONFIG_DEBUG_OBJECTS_RCU_HEAD:
check that you don't pass the same object to call_rcu()
(or friends) before an RCU grace period has elapsed
since the last time that you passed that same object to
call_rcu() (or friends).
”h]”(jY  )”}”(hŒCONFIG_DEBUG_OBJECTS_RCU_HEAD:”h]”hŒCONFIG_DEBUG_OBJECTS_RCU_HEAD:”…””}”(hj‹  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jX  hŸh³h Mêhj‡  ubji  )”}”(hhh]”hÊ)”}”(hŒ¼check that you don't pass the same object to call_rcu()
(or friends) before an RCU grace period has elapsed
since the last time that you passed that same object to
call_rcu() (or friends).”h]”hŒ¾check that you donâ€™t pass the same object to call_rcu()
(or friends) before an RCU grace period has elapsed
since the last time that you passed that same object to
call_rcu() (or friends).”…””}”(hjœ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mçhj™  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jh  hj‡  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jR  hŸh³h MêhjO  ubjS  )”}”(hŒâCONFIG_RCU_STRICT_GRACE_PERIOD:
combine with KASAN to check for pointers leaked out
of RCU read-side critical sections.  This Kconfig
option is tough on both performance and scalability,
and so is limited to four-CPU systems.
”h]”(jY  )”}”(hŒCONFIG_RCU_STRICT_GRACE_PERIOD:”h]”hŒCONFIG_RCU_STRICT_GRACE_PERIOD:”…””}”(hjº  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jX  hŸh³h Mðhj¶  ubji  )”}”(hhh]”hÊ)”}”(hŒÁcombine with KASAN to check for pointers leaked out
of RCU read-side critical sections.  This Kconfig
option is tough on both performance and scalability,
and so is limited to four-CPU systems.”h]”hŒÁcombine with KASAN to check for pointers leaked out
of RCU read-side critical sections.  This Kconfig
option is tough on both performance and scalability,
and so is limited to four-CPU systems.”…””}”(hjË  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MíhjÈ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jh  hj¶  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jR  hŸh³h MðhjO  ubjS  )”}”(hŒË__rcu sparse checks:
tag the pointer to the RCU-protected data structure
with __rcu, and sparse will warn you if you access that
pointer without the services of one of the variants
of rcu_dereference().
”h]”(jY  )”}”(hŒ__rcu sparse checks:”h]”hŒ__rcu sparse checks:”…””}”(hjé  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jX  hŸh³h Möhjå  ubji  )”}”(hhh]”hÊ)”}”(hŒµtag the pointer to the RCU-protected data structure
with __rcu, and sparse will warn you if you access that
pointer without the services of one of the variants
of rcu_dereference().”h]”hŒµtag the pointer to the RCU-protected data structure
with __rcu, and sparse will warn you if you access that
pointer without the services of one of the variants
of rcu_dereference().”…””}”(hjú  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Móhj÷  ubah}”(h]”h ]”h"]”h$]”h&]”uh1jh  hjå  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jR  hŸh³h MöhjO  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1jM  hj;  ubhÊ)”}”(hŒ_These debugging aids can help you find problems that are
otherwise extremely difficult to spot.”h]”hŒ_These debugging aids can help you find problems that are
otherwise extremely difficult to spot.”…””}”(hj  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Møhj;  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubhß)”}”(hXa  If you pass a callback function defined within a module
to one of call_rcu(), call_srcu(), call_rcu_tasks(), or
call_rcu_tasks_trace(), then it is necessary to wait for all
pending callbacks to be invoked before unloading that module.
Note that it is absolutely *not* sufficient to wait for a grace
period!  For example, synchronize_rcu() implementation is *not*
guaranteed to wait for callbacks registered on other CPUs via
call_rcu().  Or even on the current CPU if that CPU recently
went offline and came back online.

You instead need to use one of the barrier functions:

-       call_rcu() -> rcu_barrier()
-       call_srcu() -> srcu_barrier()
-       call_rcu_tasks() -> rcu_barrier_tasks()
-       call_rcu_tasks_trace() -> rcu_barrier_tasks_trace()

However, these barrier functions are absolutely *not* guaranteed
to wait for a grace period.  For example, if there are no
call_rcu() callbacks queued anywhere in the system, rcu_barrier()
can and will return immediately.

So if you need to wait for both a grace period and for all
pre-existing callbacks, you will need to invoke both functions,
with the pair depending on the flavor of RCU:

-       Either synchronize_rcu() or synchronize_rcu_expedited(),
        together with rcu_barrier()
-       Either synchronize_srcu() or synchronize_srcu_expedited(),
        together with and srcu_barrier()
-       synchronize_rcu_tasks() and rcu_barrier_tasks()
-       synchronize_tasks_trace() and rcu_barrier_tasks_trace()

If necessary, you can use something like workqueues to execute
the requisite pair of functions concurrently.

See rcubarrier.rst for more information.”h]”(hÊ)”}”(hX  If you pass a callback function defined within a module
to one of call_rcu(), call_srcu(), call_rcu_tasks(), or
call_rcu_tasks_trace(), then it is necessary to wait for all
pending callbacks to be invoked before unloading that module.
Note that it is absolutely *not* sufficient to wait for a grace
period!  For example, synchronize_rcu() implementation is *not*
guaranteed to wait for callbacks registered on other CPUs via
call_rcu().  Or even on the current CPU if that CPU recently
went offline and came back online.”h]”(hX  If you pass a callback function defined within a module
to one of call_rcu(), call_srcu(), call_rcu_tasks(), or
call_rcu_tasks_trace(), then it is necessary to wait for all
pending callbacks to be invoked before unloading that module.
Note that it is absolutely ”…””}”(hj2  hžhhŸNh NubjR  )”}”(hŒ*not*”h]”hŒnot”…””}”(hj:  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj2  ubhŒZ sufficient to wait for a grace
period!  For example, synchronize_rcu() implementation is ”…””}”(hj2  hžhhŸNh NubjR  )”}”(hŒ*not*”h]”hŒnot”…””}”(hjL  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hj2  ubhŒž
guaranteed to wait for callbacks registered on other CPUs via
call_rcu().  Or even on the current CPU if that CPU recently
went offline and came back online.”…””}”(hj2  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mûhj.  ubhÊ)”}”(hŒ5You instead need to use one of the barrier functions:”h]”hŒ5You instead need to use one of the barrier functions:”…””}”(hjd  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj.  ubhŒbullet_list”“”)”}”(hhh]”(hß)”}”(hŒcall_rcu() -> rcu_barrier()”h]”hÊ)”}”(hjy  h]”hŒcall_rcu() -> rcu_barrier()”…””}”(hj{  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhjw  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjt  ubhß)”}”(hŒcall_srcu() -> srcu_barrier()”h]”hÊ)”}”(hj  h]”hŒcall_srcu() -> srcu_barrier()”…””}”(hj’  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MhjŽ  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjt  ubhß)”}”(hŒ'call_rcu_tasks() -> rcu_barrier_tasks()”h]”hÊ)”}”(hj§  h]”hŒ'call_rcu_tasks() -> rcu_barrier_tasks()”…””}”(hj©  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M	hj¥  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjt  ubhß)”}”(hŒ4call_rcu_tasks_trace() -> rcu_barrier_tasks_trace()
”h]”hÊ)”}”(hŒ3call_rcu_tasks_trace() -> rcu_barrier_tasks_trace()”h]”hŒ3call_rcu_tasks_trace() -> rcu_barrier_tasks_trace()”…””}”(hjÀ  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h M
hj¼  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhjt  ubeh}”(h]”h ]”h"]”h$]”h&]”Œbullet”Œ-”uh1jr  hŸh³h Mhj.  ubhÊ)”}”(hŒÝHowever, these barrier functions are absolutely *not* guaranteed
to wait for a grace period.  For example, if there are no
call_rcu() callbacks queued anywhere in the system, rcu_barrier()
can and will return immediately.”h]”(hŒ0However, these barrier functions are absolutely ”…””}”(hjÜ  hžhhŸNh NubjR  )”}”(hŒ*not*”h]”hŒnot”…””}”(hjä  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1jQ  hjÜ  ubhŒ¨ guaranteed
to wait for a grace period.  For example, if there are no
call_rcu() callbacks queued anywhere in the system, rcu_barrier()
can and will return immediately.”…””}”(hjÜ  hžhhŸNh Nubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj.  ubhÊ)”}”(hŒ¨So if you need to wait for both a grace period and for all
pre-existing callbacks, you will need to invoke both functions,
with the pair depending on the flavor of RCU:”h]”hŒ¨So if you need to wait for both a grace period and for all
pre-existing callbacks, you will need to invoke both functions,
with the pair depending on the flavor of RCU:”…””}”(hjü  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj.  ubjs  )”}”(hhh]”(hß)”}”(hŒTEither synchronize_rcu() or synchronize_rcu_expedited(),
together with rcu_barrier()”h]”hÊ)”}”(hŒTEither synchronize_rcu() or synchronize_rcu_expedited(),
together with rcu_barrier()”h]”hŒTEither synchronize_rcu() or synchronize_rcu_expedited(),
together with rcu_barrier()”…””}”(hj	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj	  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj
	  ubhß)”}”(hŒ[Either synchronize_srcu() or synchronize_srcu_expedited(),
together with and srcu_barrier()”h]”hÊ)”}”(hŒ[Either synchronize_srcu() or synchronize_srcu_expedited(),
together with and srcu_barrier()”h]”hŒ[Either synchronize_srcu() or synchronize_srcu_expedited(),
together with and srcu_barrier()”…””}”(hj)	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj%	  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj
	  ubhß)”}”(hŒ/synchronize_rcu_tasks() and rcu_barrier_tasks()”h]”hÊ)”}”(hj?	  h]”hŒ/synchronize_rcu_tasks() and rcu_barrier_tasks()”…””}”(hjA	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj=	  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj
	  ubhß)”}”(hŒ8synchronize_tasks_trace() and rcu_barrier_tasks_trace()
”h]”hÊ)”}”(hŒ7synchronize_tasks_trace() and rcu_barrier_tasks_trace()”h]”hŒ7synchronize_tasks_trace() and rcu_barrier_tasks_trace()”…””}”(hjX	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h MhjT	  ubah}”(h]”h ]”h"]”h$]”h&]”uh1hÞhj
	  ubeh}”(h]”h ]”h"]”h$]”h&]”jÚ  jÛ  uh1jr  hŸh³h Mhj.  ubhÊ)”}”(hŒlIf necessary, you can use something like workqueues to execute
the requisite pair of functions concurrently.”h]”hŒlIf necessary, you can use something like workqueues to execute
the requisite pair of functions concurrently.”…””}”(hjr	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj.  ubhÊ)”}”(hŒ(See rcubarrier.rst for more information.”h]”hŒ(See rcubarrier.rst for more information.”…””}”(hj€	  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhŸh³h Mhj.  ubeh}”(h]”h ]”h"]”h$]”h&]”uh1hÞhhÛhžhhŸh³h Nubeh}”(h]”h ]”h"]”h$]”h&]”jÌ  Œarabic”jÎ  hjÏ  jÐ  Œstart”K uh1hÙhh¶hžhhŸh³h Kubeh}”(h]”Œ review-checklist-for-rcu-patches”ah ]”h"]”Œ review checklist for rcu patches”ah$]”h&]”uh1h´hhhžhhŸh³h Kubeh}”(h]”h ]”h"]”h$]”h&]”Œsource”h³uh1hŒcurrent_source”NŒcurrent_line”NŒsettings”Œdocutils.frontend”ŒValues”“”)”}”(h¹NŒ	generator”NŒ	datestamp”NŒsource_link”NŒ
source_url”NŒtoc_backlinks”Œentry”Œfootnote_backlinks”KŒsectnum_xform”KŒstrip_comments”NŒstrip_elements_with_classes”NŒstrip_classes”NŒreport_level”KŒ
halt_level”KŒexit_status_level”KŒdebug”NŒwarning_stream”NŒ	traceback”ˆŒinput_encoding”Œ	utf-8-sig”Œinput_encoding_error_handler”Œstrict”Œoutput_encoding”Œutf-8”Œoutput_encoding_error_handler”jÇ	  Œerror_encoding”Œutf-8”Œerror_encoding_error_handler”Œbackslashreplace”Œlanguage_code”Œen”Œrecord_dependencies”NŒconfig”NŒ	id_prefix”hŒauto_id_prefix”Œid”Œdump_settings”NŒdump_internals”NŒdump_transforms”NŒdump_pseudo_xml”NŒexpose_internals”NŒstrict_visitor”NŒ_disable_config”NŒ_source”h³Œ_destination”NŒ_config_files”]”Œ7/var/lib/git/docbuild/linux/Documentation/docutils.conf”aŒfile_insertion_enabled”ˆŒraw_enabled”KŒline_length_limit”M'Œpep_references”NŒpep_base_url”Œhttps://peps.python.org/”Œpep_file_url_template”Œpep-%04d”Œrfc_references”NŒrfc_base_url”Œ&https://datatracker.ietf.org/doc/html/”Œ	tab_width”KŒtrim_footnote_reference_space”‰Œsyntax_highlight”Œlong”Œsmart_quotes”ˆŒsmartquotes_locales”]”Œcharacter_level_inline_markup”‰Œdoctitle_xform”‰Œdocinfo_xform”KŒsectsubtitle_xform”‰Œimage_loading”Œlink”Œembed_stylesheet”‰Œcloak_email_addresses”ˆŒsection_self_link”‰Œenv”NubŒreporter”NŒindirect_targets”]”Œsubstitution_defs”}”Œsubstitution_names”}”Œrefnames”}”Œrefids”}”Œnameids”}”j¡	  jž	  sŒ	nametypes”}”j¡	  ‰sh}”jž	  h¶sŒfootnote_refs”}”Œcitation_refs”}”Œautofootnotes”]”Œautofootnote_refs”]”Œsymbol_footnotes”]”Œsymbol_footnote_refs”]”Œ	footnotes”]”Œ	citations”]”Œautofootnote_start”KŒsymbol_footnote_start”K Œ
id_counter”Œcollections”ŒCounter”“”}”…”R”Œparse_messages”]”hŒsystem_message”“”)”}”(hhh]”hÊ)”}”(hŒ:Enumerated list start value not ordinal-1: "0" (ordinal 0)”h]”hŒ>Enumerated list start value not ordinal-1: â€œ0â€ (ordinal 0)”…””}”(hj.
  hžhhŸNh Nubah}”(h]”h ]”h"]”h$]”h&]”uh1hÉhj+
  ubah}”(h]”h ]”h"]”h$]”h&]”Œlevel”KŒtype”ŒINFO”Œsource”h³Œline”Kuh1j)
  hh¶hžhhŸh³h KubaŒtransform_messages”]”Œtransformer”NŒinclude_log”]”Œ
decoration”Nhžhub.