hsphinx.addnodesdocument)}( rawsourcechildren]( translations LanguagesNode)}(hhh](h pending_xref)}(hhh]docutils.nodesTextChinese (Simplified)}(hhparenthuba attributes}(ids]classes]names]dupnames]backrefs] refdomainstdreftypedoc reftarget!/translations/zh_CN/RCU/whatisRCUmodnameN classnameN refexplicitutagnamehhh ubh)}(hhh]hChinese (Traditional)}(hhhh2ubah}(h]h ]h"]h$]h&] refdomainh)reftypeh+ reftarget!/translations/zh_TW/RCU/whatisRCUmodnameN classnameN refexplicituh1hhh ubh)}(hhh]hItalian}(hhhhFubah}(h]h ]h"]h$]h&] refdomainh)reftypeh+ reftarget!/translations/it_IT/RCU/whatisRCUmodnameN classnameN refexplicituh1hhh ubh)}(hhh]hJapanese}(hhhhZubah}(h]h ]h"]h$]h&] refdomainh)reftypeh+ reftarget!/translations/ja_JP/RCU/whatisRCUmodnameN classnameN refexplicituh1hhh ubh)}(hhh]hKorean}(hhhhnubah}(h]h ]h"]h$]h&] refdomainh)reftypeh+ reftarget!/translations/ko_KR/RCU/whatisRCUmodnameN classnameN refexplicituh1hhh ubh)}(hhh]hSpanish}(hhhhubah}(h]h ]h"]h$]h&] refdomainh)reftypeh+ reftarget!/translations/sp_SP/RCU/whatisRCUmodnameN classnameN refexplicituh1hhh ubeh}(h]h ]h"]h$]h&]current_languageEnglishuh1h hh _documenthsourceNlineNubhtarget)}(h.. _whatisrcu_doc:h]h}(h]h ]h"]h$]h&]refid whatisrcu-docuh1hhKhhhhh;/var/lib/git/docbuild/linux/Documentation/RCU/whatisRCU.rstubhsection)}(hhh](htitle)}(h&What is RCU? -- "Read, Copy, Update"h]h*What is RCU? -- “Read, Copy, Update”}(hhhhhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhhhhhKubh paragraph)}(haPlease note that the "What is RCU?" LWN series is an excellent place to start learning about RCU:h]hePlease note that the “What is RCU?” LWN series is an excellent place to start learning about RCU:}(hhhhhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhhhhubh line_block)}(hhh](hh)}(hC1. What is RCU, Fundamentally? https://lwn.net/Articles/262464/h](h#1. What is RCU, Fundamentally? }(h#1. What is RCU, Fundamentally? hhhhhNhNubh reference)}(h https://lwn.net/Articles/262464/h]h https://lwn.net/Articles/262464/}(hhhhhhhNhNubah}(h]h ]h"]h$]h&]refurihuh1hhhubeh}(h]h ]h"]h$]h&]uh1hindentKhhhhhhhK ubh)}(hC2. What is RCU? Part 2: Usage https://lwn.net/Articles/263130/h](h#2. What is RCU? Part 2: Usage }(h#2. What is RCU? Part 2: Usage hhhhhNhNubh)}(h https://lwn.net/Articles/263130/h]h https://lwn.net/Articles/263130/}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]refurijuh1hhhubeh}(h]h ]h"]h$]h&]uh1hhKhhhhhhhK ubh)}(hC3. RCU part 3: the RCU API https://lwn.net/Articles/264090/h](h#3. RCU part 3: the RCU API }(h#3. RCU part 3: the RCU API hjhhhNhNubh)}(h https://lwn.net/Articles/264090/h]h https://lwn.net/Articles/264090/}(hhhj#hhhNhNubah}(h]h ]h"]h$]h&]refurij%uh1hhjubeh}(h]h ]h"]h$]h&]uh1hhKhhhhhhhK ubh)}(hC4. The RCU API, 2010 Edition https://lwn.net/Articles/418853/h](h#4. The RCU API, 2010 Edition }(h#4. The RCU API, 2010 Edition hj8hhhNhNubh)}(h https://lwn.net/Articles/418853/h]h https://lwn.net/Articles/418853/}(hhhjAhhhNhNubah}(h]h ]h"]h$]h&]refurijCuh1hhj8ubeh}(h]h ]h"]h$]h&]uh1hhKhhhhhhhK ubh)}(hhh]h)}(h=2010 Big API Table https://lwn.net/Articles/419086/h](h2010 Big API Table }(h2010 Big API Table hjYhhhNhNubh)}(h https://lwn.net/Articles/419086/h]h https://lwn.net/Articles/419086/}(hhhjbhhhNhNubah}(h]h ]h"]h$]h&]refurijduh1hhjYubeh}(h]h ]h"]h$]h&]uh1hhKhjVhhhhhK ubah}(h]h ]h"]h$]h&]uh1hhhhhhhhKubh)}(hC5. The RCU API, 2014 Edition https://lwn.net/Articles/609904/h](h#5. The RCU API, 2014 Edition }(h#5. The RCU API, 2014 Edition hj}hhhNhNubh)}(h https://lwn.net/Articles/609904/h]h https://lwn.net/Articles/609904/}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]refurijuh1hhj}ubeh}(h]h ]h"]h$]h&]uh1hhKhhhhhhhKubh)}(hhh]h)}(h=2014 Big API Table https://lwn.net/Articles/609973/h](h2014 Big API Table }(h2014 Big API Table hjhhhNhNubh)}(h https://lwn.net/Articles/609973/h]h https://lwn.net/Articles/609973/}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]refurijuh1hhjubeh}(h]h ]h"]h$]h&]uh1hhKhjhhhhhKubah}(h]h ]h"]h$]h&]uh1hhhhhhhhKubh)}(hC6. The RCU API, 2019 Edition https://lwn.net/Articles/777036/h](h#6. The RCU API, 2019 Edition }(h#6. The RCU API, 2019 Edition hjhhhNhNubh)}(h https://lwn.net/Articles/777036/h]h https://lwn.net/Articles/777036/}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]refurijuh1hhjubeh}(h]h ]h"]h$]h&]uh1hhKhhhhhhhKubh)}(hhh]h)}(h=2019 Big API Table https://lwn.net/Articles/777165/h](h2019 Big API Table }(h2019 Big API Table hjhhhNhNubh)}(h https://lwn.net/Articles/777165/h]h https://lwn.net/Articles/777165/}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]refurijuh1hhjubeh}(h]h ]h"]h$]h&]uh1hhKhjhhhhhKubah}(h]h ]h"]h$]h&]uh1hhhhhhhhKubeh}(h]h ]h"]h$]h&]uh1hhhhhhhhK ubh)}(hFor those preferring video:h]hFor those preferring video:}(hjhj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhhhhubh)}(hhh](h)}(h}1. Unraveling RCU Mysteries: Fundamentals https://www.linuxfoundation.org/webinars/unraveling-rcu-usage-mysteriesh](h61. Unraveling RCU Mysteries: Fundamentals }(h61. Unraveling RCU Mysteries: Fundamentals hjhhhNhNubh)}(hGhttps://www.linuxfoundation.org/webinars/unraveling-rcu-usage-mysteriesh]hGhttps://www.linuxfoundation.org/webinars/unraveling-rcu-usage-mysteries}(hhhj'hhhNhNubah}(h]h ]h"]h$]h&]refurij)uh1hhjubeh}(h]h ]h"]h$]h&]uh1hhKhjhhhhhKubh)}(h2. Unraveling RCU Mysteries: Additional Use Cases https://www.linuxfoundation.org/webinars/unraveling-rcu-usage-mysteries-additional-use-casesh](h62. Unraveling RCU Mysteries: Additional Use Cases }(h62. Unraveling RCU Mysteries: Additional Use Cases hj<hhhNhNubh)}(h\https://www.linuxfoundation.org/webinars/unraveling-rcu-usage-mysteries-additional-use-casesh]h\https://www.linuxfoundation.org/webinars/unraveling-rcu-usage-mysteries-additional-use-cases}(hhhjEhhhNhNubah}(h]h ]h"]h$]h&]refurijGuh1hhj<ubeh}(h]h ]h"]h$]h&]uh1hhKhjhhhhhKubeh}(h]h ]h"]h$]h&]uh1hhhhhhhhKubh)}(h What is RCU?h]h What is RCU?}(hjbhj`hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhhhhubh)}(hX_RCU is a synchronization mechanism that was added to the Linux kernel during the 2.5 development effort that is optimized for read-mostly situations. Although RCU is actually quite simple, making effective use of it requires you to think differently about your code. Another part of the problem is the mistaken assumption that there is "one true way" to describe and to use RCU. Instead, the experience has been that different people must take different paths to arrive at an understanding of RCU, depending on their experiences and use cases. This document provides several different paths, as follows:h]hXcRCU is a synchronization mechanism that was added to the Linux kernel during the 2.5 development effort that is optimized for read-mostly situations. Although RCU is actually quite simple, making effective use of it requires you to think differently about your code. Another part of the problem is the mistaken assumption that there is “one true way” to describe and to use RCU. Instead, the experience has been that different people must take different paths to arrive at an understanding of RCU, depending on their experiences and use cases. This document provides several different paths, as follows:}(hjphjnhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhhhhubh)}(h+:ref:`1. RCU OVERVIEW <1_whatisRCU>`h]h)}(hj~h]hinline)}(hj~h]h1. RCU OVERVIEW}(hhhjhhhNhNubah}(h]h ](xrefstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdoc RCU/whatisRCU refdomainjreftyperef refexplicitrefwarn reftarget 1_whatisrcuuh1hhhhK%hj|ubah}(h]h ]h"]h$]h&]uh1hhhhK%hhhhubh)}(h6:ref:`2. WHAT IS RCU'S CORE API? <2_whatisRCU>`h]h)}(hjh]j)}(hjh]h#2. WHAT IS RCU’S CORE API?}(hhhjhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdocj refdomainjreftyperef refexplicitrefwarnj 2_whatisrcuuh1hhhhK'hjubah}(h]h ]h"]h$]h&]uh1hhhhK'hhhhubh)}(hJ:ref:`3. WHAT ARE SOME EXAMPLE USES OF CORE RCU API? <3_whatisRCU>`h]h)}(hjh]j)}(hjh]h53. WHAT ARE SOME EXAMPLE USES OF CORE RCU API?}(hhhjhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdocj refdomainjreftyperef refexplicitrefwarnj 3_whatisrcuuh1hhhhK)hjubah}(h]h ]h"]h$]h&]uh1hhhhK)hhhhubh)}(hG:ref:`4. WHAT IF MY UPDATING THREAD CANNOT BLOCK? <4_whatisRCU>`h]h)}(hjh]j)}(hjh]h24. WHAT IF MY UPDATING THREAD CANNOT BLOCK?}(hhhjhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdocj refdomainj reftyperef refexplicitrefwarnj 4_whatisrcuuh1hhhhK+hjubah}(h]h ]h"]h$]h&]uh1hhhhK+hhhhubh)}(hK:ref:`5. WHAT ARE SOME SIMPLE IMPLEMENTATIONS OF RCU? <5_whatisRCU>`h]h)}(hj'h]j)}(hj'h]h65. WHAT ARE SOME SIMPLE IMPLEMENTATIONS OF RCU?}(hhhj,hhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhj)ubah}(h]h ]h"]h$]h&]refdocj refdomainj6reftyperef refexplicitrefwarnj 5_whatisrcuuh1hhhhK-hj%ubah}(h]h ]h"]h$]h&]uh1hhhhK-hhhhubh)}(hA:ref:`6. ANALOGY WITH READER-WRITER LOCKING <6_whatisRCU>`h]h)}(hjPh]j)}(hjPh]h,6. ANALOGY WITH READER-WRITER LOCKING}(hhhjUhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjRubah}(h]h ]h"]h$]h&]refdocj refdomainj_reftyperef refexplicitrefwarnj 6_whatisrcuuh1hhhhK/hjNubah}(h]h ]h"]h$]h&]uh1hhhhK/hhhhubh)}(h>:ref:`7. ANALOGY WITH REFERENCE COUNTING <7_whatisRCU>`h]h)}(hjyh]j)}(hjyh]h)7. ANALOGY WITH REFERENCE COUNTING}(hhhj~hhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhj{ubah}(h]h ]h"]h$]h&]refdocj refdomainjreftyperef refexplicitrefwarnj 7_whatisrcuuh1hhhhK1hjwubah}(h]h ]h"]h$]h&]uh1hhhhK1hhhhubh)}(h4:ref:`8. FULL LIST OF RCU APIs <8_whatisRCU>`h]h)}(hjh]j)}(hjh]h8. FULL LIST OF RCU APIs}(hhhjhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdocj refdomainjreftyperef refexplicitrefwarnj 8_whatisrcuuh1hhhhK3hjubah}(h]h ]h"]h$]h&]uh1hhhhK3hhhhubh)}(h7:ref:`9. ANSWERS TO QUICK QUIZZES <9_whatisRCU>`h]h)}(hjh]j)}(hjh]h"9. ANSWERS TO QUICK QUIZZES}(hhhjhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdocj refdomainjreftyperef refexplicitrefwarnj 9_whatisrcuuh1hhhhK5hjubah}(h]h ]h"]h$]h&]uh1hhhhK5hhhhubh)}(hXvPeople who prefer starting with a conceptual overview should focus on Section 1, though most readers will profit by reading this section at some point. People who prefer to start with an API that they can then experiment with should focus on Section 2. People who prefer to start with example uses should focus on Sections 3 and 4. People who need to understand the RCU implementation should focus on Section 5, then dive into the kernel source code. People who reason best by analogy should focus on Section 6 and 7. Section 8 serves as an index to the docbook API documentation, and Section 9 is the traditional answer key.h]hXvPeople who prefer starting with a conceptual overview should focus on Section 1, though most readers will profit by reading this section at some point. People who prefer to start with an API that they can then experiment with should focus on Section 2. People who prefer to start with example uses should focus on Sections 3 and 4. People who need to understand the RCU implementation should focus on Section 5, then dive into the kernel source code. People who reason best by analogy should focus on Section 6 and 7. Section 8 serves as an index to the docbook API documentation, and Section 9 is the traditional answer key.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhK7hhhhubh)}(hX>So, start with the section that makes the most sense to you and your preferred method of learning. If you need to know everything about everything, feel free to read the whole thing -- but if you are really that type of person, you have perused the source code and will therefore never need this document anyway. ;-)h]hX>So, start with the section that makes the most sense to you and your preferred method of learning. If you need to know everything about everything, feel free to read the whole thing -- but if you are really that type of person, you have perused the source code and will therefore never need this document anyway. ;-)}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKAhhhhubh)}(h.. _1_whatisRCU:h]h}(h]h ]h"]h$]h&]h whatisrcuuh1hhKGhhhhhhubh)}(hhh](h)}(h1. RCU OVERVIEWh]h1. RCU OVERVIEW}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhjhhhhhKJubh)}(hX\The basic idea behind RCU is to split updates into "removal" and "reclamation" phases. The removal phase removes references to data items within a data structure (possibly by replacing them with references to new versions of these data items), and can run concurrently with readers. The reason that it is safe to run the removal phase concurrently with readers is the semantics of modern CPUs guarantee that readers will see either the old or the new version of the data structure rather than a partially updated reference. The reclamation phase does the work of reclaiming (e.g., freeing) the data items removed from the data structure during the removal phase. Because reclaiming data items can disrupt any readers concurrently referencing those data items, the reclamation phase must not start until readers no longer hold references to those data items.h]hXdThe basic idea behind RCU is to split updates into “removal” and “reclamation” phases. The removal phase removes references to data items within a data structure (possibly by replacing them with references to new versions of these data items), and can run concurrently with readers. The reason that it is safe to run the removal phase concurrently with readers is the semantics of modern CPUs guarantee that readers will see either the old or the new version of the data structure rather than a partially updated reference. The reclamation phase does the work of reclaiming (e.g., freeing) the data items removed from the data structure during the removal phase. Because reclaiming data items can disrupt any readers concurrently referencing those data items, the reclamation phase must not start until readers no longer hold references to those data items.}(hj,hj*hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKLhjhhubh)}(hX9Splitting the update into removal and reclamation phases permits the updater to perform the removal phase immediately, and to defer the reclamation phase until all readers active during the removal phase have completed, either by blocking until they finish or by registering a callback that is invoked after they finish. Only readers that are active during the removal phase need be considered, because any reader starting after the removal phase will be unable to gain a reference to the removed data items, and therefore cannot be disrupted by the reclamation phase.h]hX9Splitting the update into removal and reclamation phases permits the updater to perform the removal phase immediately, and to defer the reclamation phase until all readers active during the removal phase have completed, either by blocking until they finish or by registering a callback that is invoked after they finish. Only readers that are active during the removal phase need be considered, because any reader starting after the removal phase will be unable to gain a reference to the removed data items, and therefore cannot be disrupted by the reclamation phase.}(hj:hj8hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKYhjhhubh)}(hESo the typical RCU update sequence goes something like the following:h]hESo the typical RCU update sequence goes something like the following:}(hjHhjFhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKbhjhhubhenumerated_list)}(hhh](h list_item)}(h_Remove pointers to a data structure, so that subsequent readers cannot gain a reference to it. h]h)}(h^Remove pointers to a data structure, so that subsequent readers cannot gain a reference to it.h]h^Remove pointers to a data structure, so that subsequent readers cannot gain a reference to it.}(hjahj_hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKdhj[ubah}(h]h ]h"]h$]h&]uh1jYhjVhhhhhNubjZ)}(hQWait for all previous readers to complete their RCU read-side critical sections. h]h)}(hPWait for all previous readers to complete their RCU read-side critical sections.h]hPWait for all previous readers to complete their RCU read-side critical sections.}(hjyhjwhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKghjsubah}(h]h ]h"]h$]h&]uh1jYhjVhhhhhNubjZ)}(hAt this point, there cannot be any readers who hold references to the data structure, so it now may safely be reclaimed (e.g., kfree()d). h]h)}(hAt this point, there cannot be any readers who hold references to the data structure, so it now may safely be reclaimed (e.g., kfree()d).h]hAt this point, there cannot be any readers who hold references to the data structure, so it now may safely be reclaimed (e.g., kfree()d).}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKjhjubah}(h]h ]h"]h$]h&]uh1jYhjVhhhhhNubeh}(h]h ]h"]h$]h&]enumtype loweralphaprefixhsuffix.uh1jThjhhhhhKdubh)}(hXStep (b) above is the key idea underlying RCU's deferred destruction. The ability to wait until all readers are done allows RCU readers to use much lighter-weight synchronization, in some cases, absolutely no synchronization at all. In contrast, in more conventional lock-based schemes, readers must use heavy-weight synchronization in order to prevent an updater from deleting the data structure out from under them. This is because lock-based updaters typically update data items in place, and must therefore exclude readers. In contrast, RCU-based updaters typically take advantage of the fact that writes to single aligned pointers are atomic on modern CPUs, allowing atomic insertion, removal, and replacement of data items in a linked structure without disrupting readers. Concurrent RCU readers can then continue accessing the old versions, and can dispense with the atomic operations, memory barriers, and communications cache misses that are so expensive on present-day SMP computer systems, even in absence of lock contention.h]hXStep (b) above is the key idea underlying RCU’s deferred destruction. The ability to wait until all readers are done allows RCU readers to use much lighter-weight synchronization, in some cases, absolutely no synchronization at all. In contrast, in more conventional lock-based schemes, readers must use heavy-weight synchronization in order to prevent an updater from deleting the data structure out from under them. This is because lock-based updaters typically update data items in place, and must therefore exclude readers. In contrast, RCU-based updaters typically take advantage of the fact that writes to single aligned pointers are atomic on modern CPUs, allowing atomic insertion, removal, and replacement of data items in a linked structure without disrupting readers. Concurrent RCU readers can then continue accessing the old versions, and can dispense with the atomic operations, memory barriers, and communications cache misses that are so expensive on present-day SMP computer systems, even in absence of lock contention.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKnhjhhubh)}(hXaIn the three-step procedure shown above, the updater is performing both the removal and the reclamation step, but it is often helpful for an entirely different thread to do the reclamation, as is in fact the case in the Linux kernel's directory-entry cache (dcache). Even if the same thread performs both the update step (step (a) above) and the reclamation step (step (c) above), it is often helpful to think of them separately. For example, RCU readers and updaters need not communicate at all, but RCU provides implicit low-overhead communication between readers and reclaimers, namely, in step (b) above.h]hXcIn the three-step procedure shown above, the updater is performing both the removal and the reclamation step, but it is often helpful for an entirely different thread to do the reclamation, as is in fact the case in the Linux kernel’s directory-entry cache (dcache). Even if the same thread performs both the update step (step (a) above) and the reclamation step (step (c) above), it is often helpful to think of them separately. For example, RCU readers and updaters need not communicate at all, but RCU provides implicit low-overhead communication between readers and reclaimers, namely, in step (b) above.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhK~hjhhubh)}(hSo how the heck can a reclaimer tell when a reader is done, given that readers are not doing any sort of synchronization operations??? Read on to learn about how RCU's API makes this easy.h]hSo how the heck can a reclaimer tell when a reader is done, given that readers are not doing any sort of synchronization operations??? Read on to learn about how RCU’s API makes this easy.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjhhubh)}(h.. _2_whatisRCU:h]h}(h]h ]h"]h$]h&]hid1uh1hhKhjhhhhubeh}(h]( rcu-overviewjeh ]h"](1. rcu overview 1_whatisrcueh$]h&]uh1hhhhhhhhKJexpect_referenced_by_name}jjsexpect_referenced_by_id}jjsubh)}(hhh](h)}(h2. WHAT IS RCU'S CORE API?h]h2. WHAT IS RCU’S CORE API?}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhjhhhhhKubh)}(h The core RCU API is quite small:h]h The core RCU API is quite small:}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjhhubjU)}(hhh](jZ)}(hrcu_read_lock()h]h)}(hjh]hrcu_read_lock()}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjubah}(h]h ]h"]h$]h&]uh1jYhjhhhhhNubjZ)}(hrcu_read_unlock()h]h)}(hj+h]hrcu_read_unlock()}(hj+hj-hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhj)ubah}(h]h ]h"]h$]h&]uh1jYhjhhhhhNubjZ)}(hsynchronize_rcu() / call_rcu()h]h)}(hjBh]hsynchronize_rcu() / call_rcu()}(hjBhjDhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhj@ubah}(h]h ]h"]h$]h&]uh1jYhjhhhhhNubjZ)}(hrcu_assign_pointer()h]h)}(hjYh]hrcu_assign_pointer()}(hjYhj[hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjWubah}(h]h ]h"]h$]h&]uh1jYhjhhhhhNubjZ)}(hrcu_dereference() h]h)}(hrcu_dereference()h]hrcu_dereference()}(hjthjrhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjnubah}(h]h ]h"]h$]h&]uh1jYhjhhhhhNubeh}(h]h ]h"]h$]h&]jjjhjjuh1jThjhhhhhKubh)}(hThere are many other members of the RCU API, but the rest can be expressed in terms of these five, though most implementations instead express synchronize_rcu() in terms of the call_rcu() callback API.h]hThere are many other members of the RCU API, but the rest can be expressed in terms of these five, though most implementations instead express synchronize_rcu() in terms of the call_rcu() callback API.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjhhubh)}(hThe five core RCU APIs are described below, the other 18 will be enumerated later. See the kernel docbook documentation for more info, or look directly at the function header comments.h]hThe five core RCU APIs are described below, the other 18 will be enumerated later. See the kernel docbook documentation for more info, or look directly at the function header comments.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjhhubh)}(hhh](h)}(hrcu_read_lock()h]hrcu_read_lock()}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhjhhhhhKubh block_quote)}(hhh](h)}(hvoid rcu_read_lock(void);h]hvoid rcu_read_lock(void);}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjubh)}(hX5This temporal primitive is used by a reader to inform the reclaimer that the reader is entering an RCU read-side critical section. It is illegal to block while in an RCU read-side critical section, though kernels built with CONFIG_PREEMPT_RCU can preempt RCU read-side critical sections. Any RCU-protected data structure accessed during an RCU read-side critical section is guaranteed to remain unreclaimed for the full duration of that critical section. Reference counts may be used in conjunction with RCU to maintain longer-term references to data structures.h]hX5This temporal primitive is used by a reader to inform the reclaimer that the reader is entering an RCU read-side critical section. It is illegal to block while in an RCU read-side critical section, though kernels built with CONFIG_PREEMPT_RCU can preempt RCU read-side critical sections. Any RCU-protected data structure accessed during an RCU read-side critical section is guaranteed to remain unreclaimed for the full duration of that critical section. Reference counts may be used in conjunction with RCU to maintain longer-term references to data structures.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjubh)}(hXeNote that anything that disables bottom halves, preemption, or interrupts also enters an RCU read-side critical section. Acquiring a spinlock also enters an RCU read-side critical sections, even for spinlocks that do not disable preemption, as is the case in kernels built with CONFIG_PREEMPT_RT=y. Sleeplocks do *not* enter RCU read-side critical sections.h](hX9Note that anything that disables bottom halves, preemption, or interrupts also enters an RCU read-side critical section. Acquiring a spinlock also enters an RCU read-side critical sections, even for spinlocks that do not disable preemption, as is the case in kernels built with CONFIG_PREEMPT_RT=y. Sleeplocks do }(hX9Note that anything that disables bottom halves, preemption, or interrupts also enters an RCU read-side critical section. Acquiring a spinlock also enters an RCU read-side critical sections, even for spinlocks that do not disable preemption, as is the case in kernels built with CONFIG_PREEMPT_RT=y. Sleeplocks do hjhhhNhNubhemphasis)}(h*not*h]hnot}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]uh1jhjubh' enter RCU read-side critical sections.}(h' enter RCU read-side critical sections.hjhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhKhjubeh}(h]h ]h"]h$]h&]uh1jhjhhhhhNubeh}(h] rcu-read-lockah ]h"]rcu_read_lock()ah$]h&]uh1hhjhhhhhKubh)}(hhh](h)}(hrcu_read_unlock()h]hrcu_read_unlock()}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhj hhhhhKubj)}(hhh](h)}(hvoid rcu_read_unlock(void);h]hvoid rcu_read_unlock(void);}(hj"hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjubh)}(hX3This temporal primitives is used by a reader to inform the reclaimer that the reader is exiting an RCU read-side critical section. Anything that enables bottom halves, preemption, or interrupts also exits an RCU read-side critical section. Releasing a spinlock also exits an RCU read-side critical section.h]hX3This temporal primitives is used by a reader to inform the reclaimer that the reader is exiting an RCU read-side critical section. Anything that enables bottom halves, preemption, or interrupts also exits an RCU read-side critical section. Releasing a spinlock also exits an RCU read-side critical section.}(hj0hj.hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjubh)}(hKNote that RCU read-side critical sections may be nested and/or overlapping.h]hKNote that RCU read-side critical sections may be nested and/or overlapping.}(hj>hj<hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjubeh}(h]h ]h"]h$]h&]uh1jhj hhhhhNubeh}(h]rcu-read-unlockah ]h"]rcu_read_unlock()ah$]h&]uh1hhjhhhhhKubh)}(hhh](h)}(hsynchronize_rcu()h]hsynchronize_rcu()}(hj]hj[hhhNhNubah}(h]h ]h"]h$]h&]uh1hhjXhhhhhKubj)}(hhh](h)}(hvoid synchronize_rcu(void);h]hvoid synchronize_rcu(void);}(hjnhjlhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjiubh)}(hX|This temporal primitive marks the end of updater code and the beginning of reclaimer code. It does this by blocking until all pre-existing RCU read-side critical sections on all CPUs have completed. Note that synchronize_rcu() will **not** necessarily wait for any subsequent RCU read-side critical sections to complete. For example, consider the following sequence of events::h](hThis temporal primitive marks the end of updater code and the beginning of reclaimer code. It does this by blocking until all pre-existing RCU read-side critical sections on all CPUs have completed. Note that synchronize_rcu() will }(hThis temporal primitive marks the end of updater code and the beginning of reclaimer code. It does this by blocking until all pre-existing RCU read-side critical sections on all CPUs have completed. Note that synchronize_rcu() will hjzhhhNhNubhstrong)}(h**not**h]hnot}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]uh1jhjzubh necessarily wait for any subsequent RCU read-side critical sections to complete. For example, consider the following sequence of events:}(h necessarily wait for any subsequent RCU read-side critical sections to complete. For example, consider the following sequence of events:hjzhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhKhjiubh literal_block)}(hX CPU 0 CPU 1 CPU 2 ----------------- ------------------------- --------------- 1. rcu_read_lock() 2. enters synchronize_rcu() 3. rcu_read_lock() 4. rcu_read_unlock() 5. exits synchronize_rcu() 6. rcu_read_unlock()h]hX CPU 0 CPU 1 CPU 2 ----------------- ------------------------- --------------- 1. rcu_read_lock() 2. enters synchronize_rcu() 3. rcu_read_lock() 4. rcu_read_unlock() 5. exits synchronize_rcu() 6. rcu_read_unlock()}(hhhjubah}(h]h ]h"]h$]h&] xml:spacepreserveuh1jhhhKhjiubh)}(hTo reiterate, synchronize_rcu() waits only for ongoing RCU read-side critical sections to complete, not necessarily for any that begin after synchronize_rcu() is invoked.h]hTo reiterate, synchronize_rcu() waits only for ongoing RCU read-side critical sections to complete, not necessarily for any that begin after synchronize_rcu() is invoked.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjiubh)}(hX[Of course, synchronize_rcu() does not necessarily return **immediately** after the last pre-existing RCU read-side critical section completes. For one thing, there might well be scheduling delays. For another thing, many RCU implementations process requests in batches in order to improve efficiencies, which can further delay synchronize_rcu().h](h9Of course, synchronize_rcu() does not necessarily return }(h9Of course, synchronize_rcu() does not necessarily return hjhhhNhNubj)}(h**immediately**h]h immediately}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]uh1jhjubhX after the last pre-existing RCU read-side critical section completes. For one thing, there might well be scheduling delays. For another thing, many RCU implementations process requests in batches in order to improve efficiencies, which can further delay synchronize_rcu().}(hX after the last pre-existing RCU read-side critical section completes. For one thing, there might well be scheduling delays. For another thing, many RCU implementations process requests in batches in order to improve efficiencies, which can further delay synchronize_rcu().hjhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhKhjiubh)}(hSince synchronize_rcu() is the API that must figure out when readers are done, its implementation is key to RCU. For RCU to be useful in all but the most read-intensive situations, synchronize_rcu()'s overhead must also be quite small.h]hSince synchronize_rcu() is the API that must figure out when readers are done, its implementation is key to RCU. For RCU to be useful in all but the most read-intensive situations, synchronize_rcu()’s overhead must also be quite small.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjiubh)}(hXThe call_rcu() API is an asynchronous callback form of synchronize_rcu(), and is described in more detail in a later section. Instead of blocking, it registers a function and argument which are invoked after all ongoing RCU read-side critical sections have completed. This callback variant is particularly useful in situations where it is illegal to block or where update-side performance is critically important.h]hXThe call_rcu() API is an asynchronous callback form of synchronize_rcu(), and is described in more detail in a later section. Instead of blocking, it registers a function and argument which are invoked after all ongoing RCU read-side critical sections have completed. This callback variant is particularly useful in situations where it is illegal to block or where update-side performance is critically important.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjiubh)}(hXHowever, the call_rcu() API should not be used lightly, as use of the synchronize_rcu() API generally results in simpler code. In addition, the synchronize_rcu() API has the nice property of automatically limiting update rate should grace periods be delayed. This property results in system resilience in face of denial-of-service attacks. Code using call_rcu() should limit update rate in order to gain this same sort of resilience. See checklist.rst for some approaches to limiting the update rate.h]hXHowever, the call_rcu() API should not be used lightly, as use of the synchronize_rcu() API generally results in simpler code. In addition, the synchronize_rcu() API has the nice property of automatically limiting update rate should grace periods be delayed. This property results in system resilience in face of denial-of-service attacks. Code using call_rcu() should limit update rate in order to gain this same sort of resilience. See checklist.rst for some approaches to limiting the update rate.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhjiubeh}(h]h ]h"]h$]h&]uh1jhjXhhhhhNubeh}(h]synchronize-rcuah ]h"]synchronize_rcu()ah$]h&]uh1hhjhhhhhKubh)}(hhh](h)}(hrcu_assign_pointer()h]hrcu_assign_pointer()}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhjhhhhhKubj)}(hhh](h)}(h(void rcu_assign_pointer(p, typeof(p) v);h]h(void rcu_assign_pointer(p, typeof(p) v);}(hj.hj,hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhKhj)ubh)}(hYes, rcu_assign_pointer() **is** implemented as a macro, though it would be cool to be able to declare a function in this manner. (Compiler experts will no doubt disagree.)h](hYes, rcu_assign_pointer() }(hYes, rcu_assign_pointer() hj:hhhNhNubj)}(h**is**h]his}(hhhjChhhNhNubah}(h]h ]h"]h$]h&]uh1jhj:ubh implemented as a macro, though it would be cool to be able to declare a function in this manner. (Compiler experts will no doubt disagree.)}(h implemented as a macro, though it would be cool to be able to declare a function in this manner. (Compiler experts will no doubt disagree.)hj:hhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhKhj)ubh)}(hXThe updater uses this spatial macro to assign a new value to an RCU-protected pointer, in order to safely communicate the change in value from the updater to the reader. This is a spatial (as opposed to temporal) macro. It does not evaluate to an rvalue, but it does execute any memory-barrier instructions required for a given CPU architecture. Its ordering properties are that of a store-release operation.h]hXThe updater uses this spatial macro to assign a new value to an RCU-protected pointer, in order to safely communicate the change in value from the updater to the reader. This is a spatial (as opposed to temporal) macro. It does not evaluate to an rvalue, but it does execute any memory-barrier instructions required for a given CPU architecture. Its ordering properties are that of a store-release operation.}(hj^hj\hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj)ubh)}(hX.Perhaps just as important, it serves to document (1) which pointers are protected by RCU and (2) the point at which a given structure becomes accessible to other CPUs. That said, rcu_assign_pointer() is most frequently used indirectly, via the _rcu list-manipulation primitives such as list_add_rcu().h]hX.Perhaps just as important, it serves to document (1) which pointers are protected by RCU and (2) the point at which a given structure becomes accessible to other CPUs. That said, rcu_assign_pointer() is most frequently used indirectly, via the _rcu list-manipulation primitives such as list_add_rcu().}(hjlhjjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM hj)ubeh}(h]h ]h"]h$]h&]uh1jhjhhhhhNubeh}(h]rcu-assign-pointerah ]h"]rcu_assign_pointer()ah$]h&]uh1hhjhhhhhKubh)}(hhh](h)}(hrcu_dereference()h]hrcu_dereference()}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhjhhhhhMubj)}(hhh](h)}(htypeof(p) rcu_dereference(p);h]htypeof(p) rcu_dereference(p);}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubh)}(hLLike rcu_assign_pointer(), rcu_dereference() must be implemented as a macro.h]hLLike rcu_assign_pointer(), rcu_dereference() must be implemented as a macro.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubh)}(hXThe reader uses the spatial rcu_dereference() macro to fetch an RCU-protected pointer, which returns a value that may then be safely dereferenced. Note that rcu_dereference() does not actually dereference the pointer, instead, it protects the pointer for later dereferencing. It also executes any needed memory-barrier instructions for a given CPU architecture. Currently, only Alpha needs memory barriers within rcu_dereference() -- on other CPUs, it compiles to a volatile load.h]hXThe reader uses the spatial rcu_dereference() macro to fetch an RCU-protected pointer, which returns a value that may then be safely dereferenced. Note that rcu_dereference() does not actually dereference the pointer, instead, it protects the pointer for later dereferencing. It also executes any needed memory-barrier instructions for a given CPU architecture. Currently, only Alpha needs memory barriers within rcu_dereference() -- on other CPUs, it compiles to a volatile load.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubh)}(hCommon coding practice uses rcu_dereference() to copy an RCU-protected pointer to a local variable, then dereferences this local variable, for example as follows::h]hCommon coding practice uses rcu_dereference() to copy an RCU-protected pointer to a local variable, then dereferences this local variable, for example as follows:}(hCommon coding practice uses rcu_dereference() to copy an RCU-protected pointer to a local variable, then dereferences this local variable, for example as follows:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM hjubj)}(h/p = rcu_dereference(head.next); return p->data;h]h/p = rcu_dereference(head.next); return p->data;}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhM$hjubh)}(hRHowever, in this case, one could just as easily combine these into one statement::h]hQHowever, in this case, one could just as easily combine these into one statement:}(hQHowever, in this case, one could just as easily combine these into one statement:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM'hjubj)}(h(return rcu_dereference(head.next)->data;h]h(return rcu_dereference(head.next)->data;}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhM*hjubh)}(hXJIf you are going to be fetching multiple fields from the RCU-protected structure, using the local variable is of course preferred. Repeated rcu_dereference() calls look ugly, do not guarantee that the same pointer will be returned if an update happened while in the critical section, and incur unnecessary overhead on Alpha CPUs.h]hXJIf you are going to be fetching multiple fields from the RCU-protected structure, using the local variable is of course preferred. Repeated rcu_dereference() calls look ugly, do not guarantee that the same pointer will be returned if an update happened while in the critical section, and incur unnecessary overhead on Alpha CPUs.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM,hjubh)}(hNote that the value returned by rcu_dereference() is valid only within the enclosing RCU read-side critical section [1]_. For example, the following is **not** legal::h](htNote that the value returned by rcu_dereference() is valid only within the enclosing RCU read-side critical section }(htNote that the value returned by rcu_dereference() is valid only within the enclosing RCU read-side critical section hj hhhNhNubhfootnote_reference)}(h[1]_h]h1}(hhhjhhhNhNubah}(h]id2ah ]h"]h$]h&]hid4docnamejuh1jhj resolvedKubh . For example, the following is }(h . For example, the following is hj hhhNhNubj)}(h**not**h]hnot}(hhhj.hhhNhNubah}(h]h ]h"]h$]h&]uh1jhj ubh legal:}(h legal:hj hhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhM3hjubj)}(hrcu_read_lock(); p = rcu_dereference(head.next); rcu_read_unlock(); x = p->address; /* BUG!!! */ rcu_read_lock(); y = p->data; /* BUG!!! */ rcu_read_unlock();h]hrcu_read_lock(); p = rcu_dereference(head.next); rcu_read_unlock(); x = p->address; /* BUG!!! */ rcu_read_lock(); y = p->data; /* BUG!!! */ rcu_read_unlock();}(hhhjGubah}(h]h ]h"]h$]h&]jjuh1jhhhM7hjubh)}(hX,Holding a reference from one RCU read-side critical section to another is just as illegal as holding a reference from one lock-based critical section to another! Similarly, using a reference outside of the critical section in which it was acquired is just as illegal as doing so with normal locking.h]hX,Holding a reference from one RCU read-side critical section to another is just as illegal as holding a reference from one lock-based critical section to another! Similarly, using a reference outside of the critical section in which it was acquired is just as illegal as doing so with normal locking.}(hjWhjUhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM?hjubh)}(hXAs with rcu_assign_pointer(), an important function of rcu_dereference() is to document which pointers are protected by RCU, in particular, flagging a pointer that is subject to changing at any time, including immediately after the rcu_dereference(). And, again like rcu_assign_pointer(), rcu_dereference() is typically used indirectly, via the _rcu list-manipulation primitives, such as list_for_each_entry_rcu() [2]_.h](hXAs with rcu_assign_pointer(), an important function of rcu_dereference() is to document which pointers are protected by RCU, in particular, flagging a pointer that is subject to changing at any time, including immediately after the rcu_dereference(). And, again like rcu_assign_pointer(), rcu_dereference() is typically used indirectly, via the _rcu list-manipulation primitives, such as list_for_each_entry_rcu() }(hXAs with rcu_assign_pointer(), an important function of rcu_dereference() is to document which pointers are protected by RCU, in particular, flagging a pointer that is subject to changing at any time, including immediately after the rcu_dereference(). And, again like rcu_assign_pointer(), rcu_dereference() is typically used indirectly, via the _rcu list-manipulation primitives, such as list_for_each_entry_rcu() hjchhhNhNubj)}(h[2]_h]h2}(hhhjlhhhNhNubah}(h]id3ah ]h"]h$]h&]hid5j'juh1jhjcj(Kubh.}(hjhjchhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMFhjubeh}(h]h ]h"]h$]h&]uh1jhjhhhhhNubhfootnote)}(hXThe variant rcu_dereference_protected() can be used outside of an RCU read-side critical section as long as the usage is protected by locks acquired by the update-side code. This variant avoids the lockdep warning that would happen when using (for example) rcu_dereference() without rcu_read_lock() protection. Using rcu_dereference_protected() also has the advantage of permitting compiler optimizations that rcu_dereference() must prohibit. The rcu_dereference_protected() variant takes a lockdep expression to indicate which locks must be acquired by the caller. If the indicated protection is not provided, a lockdep splat is emitted. See Design/Requirements/Requirements.rst and the API's code comments for more details and example usage. h](hlabel)}(h1h]h1}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]uh1jhjubh)}(hXThe variant rcu_dereference_protected() can be used outside of an RCU read-side critical section as long as the usage is protected by locks acquired by the update-side code. This variant avoids the lockdep warning that would happen when using (for example) rcu_dereference() without rcu_read_lock() protection. Using rcu_dereference_protected() also has the advantage of permitting compiler optimizations that rcu_dereference() must prohibit. The rcu_dereference_protected() variant takes a lockdep expression to indicate which locks must be acquired by the caller. If the indicated protection is not provided, a lockdep splat is emitted. See Design/Requirements/Requirements.rst and the API's code comments for more details and example usage.h]hXThe variant rcu_dereference_protected() can be used outside of an RCU read-side critical section as long as the usage is protected by locks acquired by the update-side code. This variant avoids the lockdep warning that would happen when using (for example) rcu_dereference() without rcu_read_lock() protection. Using rcu_dereference_protected() also has the advantage of permitting compiler optimizations that rcu_dereference() must prohibit. The rcu_dereference_protected() variant takes a lockdep expression to indicate which locks must be acquired by the caller. If the indicated protection is not provided, a lockdep splat is emitted. See Design/Requirements/Requirements.rst and the API’s code comments for more details and example usage.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMNhjubeh}(h]j&ah ]h"]1ah$]h&]j!aj'juh1jhhhMNhjhhj(Kubj)}(hXIf the list_for_each_entry_rcu() instance might be used by update-side code as well as by RCU readers, then an additional lockdep expression can be added to its list of arguments. For example, given an additional "lock_is_held(&mylock)" argument, the RCU lockdep code would complain only if this instance was invoked outside of an RCU read-side critical section and without the protection of mylock. h](j)}(h2h]h2}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]uh1jhjubh)}(hXIf the list_for_each_entry_rcu() instance might be used by update-side code as well as by RCU readers, then an additional lockdep expression can be added to its list of arguments. For example, given an additional "lock_is_held(&mylock)" argument, the RCU lockdep code would complain only if this instance was invoked outside of an RCU read-side critical section and without the protection of mylock.h]hXIf the list_for_each_entry_rcu() instance might be used by update-side code as well as by RCU readers, then an additional lockdep expression can be added to its list of arguments. For example, given an additional “lock_is_held(&mylock)” argument, the RCU lockdep code would complain only if this instance was invoked outside of an RCU read-side critical section and without the protection of mylock.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM[hjubeh}(h]j{ah ]h"]2ah$]h&]jvaj'juh1jhhhM[hjhhj(Kubh)}(hbThe following diagram shows how each API communicates among the reader, updater, and reclaimer. ::h]h_The following diagram shows how each API communicates among the reader, updater, and reclaimer.}(h_The following diagram shows how each API communicates among the reader, updater, and reclaimer.hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMchjhhubj)}(hXrcu_assign_pointer() +--------+ +---------------------->| reader |---------+ | +--------+ | | | | | | | Protect: | | | rcu_read_lock() | | | rcu_read_unlock() | rcu_dereference() | | +---------+ | | | updater |<----------------+ | +---------+ V | +-----------+ +----------------------------------->| reclaimer | +-----------+ Defer: synchronize_rcu() & call_rcu()h]hXrcu_assign_pointer() +--------+ +---------------------->| reader |---------+ | +--------+ | | | | | | | Protect: | | | rcu_read_lock() | | | rcu_read_unlock() | rcu_dereference() | | +---------+ | | | updater |<----------------+ | +---------+ V | +-----------+ +----------------------------------->| reclaimer | +-----------+ Defer: synchronize_rcu() & call_rcu()}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMhhjhhubh)}(hXIThe RCU infrastructure observes the temporal sequence of rcu_read_lock(), rcu_read_unlock(), synchronize_rcu(), and call_rcu() invocations in order to determine when (1) synchronize_rcu() invocations may return to their callers and (2) call_rcu() callbacks may be invoked. Efficient implementations of the RCU infrastructure make heavy use of batching in order to amortize their overhead over many uses of the corresponding APIs. The rcu_assign_pointer() and rcu_dereference() invocations communicate spatial changes via stores to and loads from the RCU-protected pointer in question.h]hXIThe RCU infrastructure observes the temporal sequence of rcu_read_lock(), rcu_read_unlock(), synchronize_rcu(), and call_rcu() invocations in order to determine when (1) synchronize_rcu() invocations may return to their callers and (2) call_rcu() callbacks may be invoked. Efficient implementations of the RCU infrastructure make heavy use of batching in order to amortize their overhead over many uses of the corresponding APIs. The rcu_assign_pointer() and rcu_dereference() invocations communicate spatial changes via stores to and loads from the RCU-protected pointer in question.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM{hjhhubh)}(hXRThere are at least three flavors of RCU usage in the Linux kernel. The diagram above shows the most common one. On the updater side, the rcu_assign_pointer(), synchronize_rcu() and call_rcu() primitives used are the same for all three flavors. However for protection (on the reader side), the primitives used vary depending on the flavor:h]hXRThere are at least three flavors of RCU usage in the Linux kernel. The diagram above shows the most common one. On the updater side, the rcu_assign_pointer(), synchronize_rcu() and call_rcu() primitives used are the same for all three flavors. However for protection (on the reader side), the primitives used vary depending on the flavor:}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubjU)}(hhh](jZ)}(h6rcu_read_lock() / rcu_read_unlock() rcu_dereference() h]h)}(h5rcu_read_lock() / rcu_read_unlock() rcu_dereference()h]h5rcu_read_lock() / rcu_read_unlock() rcu_dereference()}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj ubah}(h]h ]h"]h$]h&]uh1jYhj hhhhhNubjZ)}(hfrcu_read_lock_bh() / rcu_read_unlock_bh() local_bh_disable() / local_bh_enable() rcu_dereference_bh() h]h)}(hercu_read_lock_bh() / rcu_read_unlock_bh() local_bh_disable() / local_bh_enable() rcu_dereference_bh()h]hercu_read_lock_bh() / rcu_read_unlock_bh() local_bh_disable() / local_bh_enable() rcu_dereference_bh()}(hj8 hj6 hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj2 ubah}(h]h ]h"]h$]h&]uh1jYhj hhhhhNubjZ)}(hrcu_read_lock_sched() / rcu_read_unlock_sched() preempt_disable() / preempt_enable() local_irq_save() / local_irq_restore() hardirq enter / hardirq exit NMI enter / NMI exit rcu_dereference_sched() h]h)}(hrcu_read_lock_sched() / rcu_read_unlock_sched() preempt_disable() / preempt_enable() local_irq_save() / local_irq_restore() hardirq enter / hardirq exit NMI enter / NMI exit rcu_dereference_sched()h]hrcu_read_lock_sched() / rcu_read_unlock_sched() preempt_disable() / preempt_enable() local_irq_save() / local_irq_restore() hardirq enter / hardirq exit NMI enter / NMI exit rcu_dereference_sched()}(hjP hjN hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjJ ubah}(h]h ]h"]h$]h&]uh1jYhj hhhhhNubeh}(h]h ]h"]h$]h&]jjjhjjuh1jThjhhhhhMubh)}(h(These three flavors are used as follows:h]h(These three flavors are used as follows:}(hjj hjh hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubjU)}(hhh](jZ)}(h'RCU applied to normal data structures. h]h)}(h&RCU applied to normal data structures.h]h&RCU applied to normal data structures.}(hj hj} hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjy ubah}(h]h ]h"]h$]h&]uh1jYhjv hhhhhNubjZ)}(heRCU applied to networking data structures that may be subjected to remote denial-of-service attacks. h]h)}(hdRCU applied to networking data structures that may be subjected to remote denial-of-service attacks.h]hdRCU applied to networking data structures that may be subjected to remote denial-of-service attacks.}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj ubah}(h]h ]h"]h$]h&]uh1jYhjv hhhhhNubjZ)}(h:RCU applied to scheduler and interrupt/NMI-handler tasks. h]h)}(h9RCU applied to scheduler and interrupt/NMI-handler tasks.h]h9RCU applied to scheduler and interrupt/NMI-handler tasks.}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj ubah}(h]h ]h"]h$]h&]uh1jYhjv hhhhhNubeh}(h]h ]h"]h$]h&]jjjhjjuh1jThjhhhhhMubh)}(hAgain, most uses will be of (a). The (b) and (c) cases are important for specialized uses, but are relatively uncommon. The SRCU, RCU-Tasks, RCU-Tasks-Rude, and RCU-Tasks-Trace have similar relationships among their assorted primitives.h]hAgain, most uses will be of (a). The (b) and (c) cases are important for specialized uses, but are relatively uncommon. The SRCU, RCU-Tasks, RCU-Tasks-Rude, and RCU-Tasks-Trace have similar relationships among their assorted primitives.}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubh)}(h.. _3_whatisRCU:h]h}(h]h ]h"]h$]h&]hid6uh1hhMhjhhhhubeh}(h]rcu-dereferenceah ]h"]rcu_dereference()ah$]h&]uh1hhjhhhhhMubeh}(h](what-is-rcu-s-core-apijeh ]h"](2. what is rcu's core api? 2_whatisrcueh$]h&]uh1hhhhhhhhKj}j jsj}jjsubh)}(hhh](h)}(h/3. WHAT ARE SOME EXAMPLE USES OF CORE RCU API?h]h/3. WHAT ARE SOME EXAMPLE USES OF CORE RCU API?}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhj hhhhhMubh)}(hThis section shows a simple use of the core RCU API to protect a global pointer to a dynamically allocated structure. More-typical uses of RCU may be found in listRCU.rst and NMI-RCU.rst. ::h]hThis section shows a simple use of the core RCU API to protect a global pointer to a dynamically allocated structure. More-typical uses of RCU may be found in listRCU.rst and NMI-RCU.rst.}(hThis section shows a simple use of the core RCU API to protect a global pointer to a dynamically allocated structure. More-typical uses of RCU may be found in listRCU.rst and NMI-RCU.rst.hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubj)}(hX6struct foo { int a; char b; long c; }; DEFINE_SPINLOCK(foo_mutex); struct foo __rcu *gbl_foo; /* * Create a new struct foo that is the same as the one currently * pointed to by gbl_foo, except that field "a" is replaced * with "new_a". Points gbl_foo to the new structure, and * frees up the old structure after a grace period. * * Uses rcu_assign_pointer() to ensure that concurrent readers * see the initialized version of the new structure. * * Uses synchronize_rcu() to ensure that any readers that might * have references to the old structure complete before freeing * the old structure. */ void foo_update_a(int new_a) { struct foo *new_fp; struct foo *old_fp; new_fp = kmalloc(sizeof(*new_fp), GFP_KERNEL); spin_lock(&foo_mutex); old_fp = rcu_dereference_protected(gbl_foo, lockdep_is_held(&foo_mutex)); *new_fp = *old_fp; new_fp->a = new_a; rcu_assign_pointer(gbl_foo, new_fp); spin_unlock(&foo_mutex); synchronize_rcu(); kfree(old_fp); } /* * Return the value of field "a" of the current gbl_foo * structure. Use rcu_read_lock() and rcu_read_unlock() * to ensure that the structure does not get deleted out * from under us, and use rcu_dereference() to ensure that * we see the initialized version of the structure (important * for DEC Alpha and for people reading the code). */ int foo_get_a(void) { int retval; rcu_read_lock(); retval = rcu_dereference(gbl_foo)->a; rcu_read_unlock(); return retval; }h]hX6struct foo { int a; char b; long c; }; DEFINE_SPINLOCK(foo_mutex); struct foo __rcu *gbl_foo; /* * Create a new struct foo that is the same as the one currently * pointed to by gbl_foo, except that field "a" is replaced * with "new_a". Points gbl_foo to the new structure, and * frees up the old structure after a grace period. * * Uses rcu_assign_pointer() to ensure that concurrent readers * see the initialized version of the new structure. * * Uses synchronize_rcu() to ensure that any readers that might * have references to the old structure complete before freeing * the old structure. */ void foo_update_a(int new_a) { struct foo *new_fp; struct foo *old_fp; new_fp = kmalloc(sizeof(*new_fp), GFP_KERNEL); spin_lock(&foo_mutex); old_fp = rcu_dereference_protected(gbl_foo, lockdep_is_held(&foo_mutex)); *new_fp = *old_fp; new_fp->a = new_a; rcu_assign_pointer(gbl_foo, new_fp); spin_unlock(&foo_mutex); synchronize_rcu(); kfree(old_fp); } /* * Return the value of field "a" of the current gbl_foo * structure. Use rcu_read_lock() and rcu_read_unlock() * to ensure that the structure does not get deleted out * from under us, and use rcu_dereference() to ensure that * we see the initialized version of the structure (important * for DEC Alpha and for people reading the code). */ int foo_get_a(void) { int retval; rcu_read_lock(); retval = rcu_dereference(gbl_foo)->a; rcu_read_unlock(); return retval; }}(hhhj ubah}(h]h ]h"]h$]h&]jjuh1jhhhMhj hhubh)}(hSo, to sum up:h]hSo, to sum up:}(hj# hj! hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh bullet_list)}(hhh](jZ)}(hTUse rcu_read_lock() and rcu_read_unlock() to guard RCU read-side critical sections. h]h)}(hSUse rcu_read_lock() and rcu_read_unlock() to guard RCU read-side critical sections.h]hSUse rcu_read_lock() and rcu_read_unlock() to guard RCU read-side critical sections.}(hj: hj8 hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj4 ubah}(h]h ]h"]h$]h&]uh1jYhj1 hhhhhNubjZ)}(hgWithin an RCU read-side critical section, use rcu_dereference() to dereference RCU-protected pointers. h]h)}(hfWithin an RCU read-side critical section, use rcu_dereference() to dereference RCU-protected pointers.h]hfWithin an RCU read-side critical section, use rcu_dereference() to dereference RCU-protected pointers.}(hjR hjP hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjL ubah}(h]h ]h"]h$]h&]uh1jYhj1 hhhhhNubjZ)}(hqUse some solid design (such as locks or semaphores) to keep concurrent updates from interfering with each other. h]h)}(hpUse some solid design (such as locks or semaphores) to keep concurrent updates from interfering with each other.h]hpUse some solid design (such as locks or semaphores) to keep concurrent updates from interfering with each other.}(hjj hjh hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjd ubah}(h]h ]h"]h$]h&]uh1jYhj1 hhhhhNubjZ)}(hX;Use rcu_assign_pointer() to update an RCU-protected pointer. This primitive protects concurrent readers from the updater, **not** concurrent updates from each other! You therefore still need to use locking (or something similar) to keep concurrent rcu_assign_pointer() primitives from interfering with each other. h]h)}(hX:Use rcu_assign_pointer() to update an RCU-protected pointer. This primitive protects concurrent readers from the updater, **not** concurrent updates from each other! You therefore still need to use locking (or something similar) to keep concurrent rcu_assign_pointer() primitives from interfering with each other.h](hzUse rcu_assign_pointer() to update an RCU-protected pointer. This primitive protects concurrent readers from the updater, }(hzUse rcu_assign_pointer() to update an RCU-protected pointer. This primitive protects concurrent readers from the updater, hj hhhNhNubj)}(h**not**h]hnot}(hhhj hhhNhNubah}(h]h ]h"]h$]h&]uh1jhj ubh concurrent updates from each other! You therefore still need to use locking (or something similar) to keep concurrent rcu_assign_pointer() primitives from interfering with each other.}(h concurrent updates from each other! You therefore still need to use locking (or something similar) to keep concurrent rcu_assign_pointer() primitives from interfering with each other.hj hhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMhj| ubah}(h]h ]h"]h$]h&]uh1jYhj1 hhhhhNubjZ)}(hXUse synchronize_rcu() **after** removing a data element from an RCU-protected data structure, but **before** reclaiming/freeing the data element, in order to wait for the completion of all RCU read-side critical sections that might be referencing that data item. h]h)}(hXUse synchronize_rcu() **after** removing a data element from an RCU-protected data structure, but **before** reclaiming/freeing the data element, in order to wait for the completion of all RCU read-side critical sections that might be referencing that data item.h](hUse synchronize_rcu() }(hUse synchronize_rcu() hj hhhNhNubj)}(h **after**h]hafter}(hhhj hhhNhNubah}(h]h ]h"]h$]h&]uh1jhj ubhC removing a data element from an RCU-protected data structure, but }(hC removing a data element from an RCU-protected data structure, but hj hhhNhNubj)}(h **before**h]hbefore}(hhhj hhhNhNubah}(h]h ]h"]h$]h&]uh1jhj ubh reclaiming/freeing the data element, in order to wait for the completion of all RCU read-side critical sections that might be referencing that data item.}(h reclaiming/freeing the data element, in order to wait for the completion of all RCU read-side critical sections that might be referencing that data item.hj hhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMhj ubah}(h]h ]h"]h$]h&]uh1jYhj1 hhhhhNubeh}(h]h ]h"]h$]h&]bullet-uh1j/ hhhMhj hhubh)}(hSee checklist.rst for additional rules to follow when using RCU. And again, more-typical uses of RCU may be found in listRCU.rst and NMI-RCU.rst.h]hSee checklist.rst for additional rules to follow when using RCU. And again, more-typical uses of RCU may be found in listRCU.rst and NMI-RCU.rst.}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(h.. _4_whatisRCU:h]h}(h]h ]h"]h$]h&]hid7uh1hhMhj hhhhubeh}(h](*what-are-some-example-uses-of-core-rcu-apij eh ]h"](.3. what are some example uses of core rcu api? 3_whatisrcueh$]h&]uh1hhhhhhhhMj}j j sj}j j subh)}(hhh](h)}(h,4. WHAT IF MY UPDATING THREAD CANNOT BLOCK?h]h,4. WHAT IF MY UPDATING THREAD CANNOT BLOCK?}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhj hhhhhMubh)}(hIn the example above, foo_update_a() blocks until a grace period elapses. This is quite simple, but in some cases one cannot afford to wait so long -- there might be other high-priority work to be done.h]hIn the example above, foo_update_a() blocks until a grace period elapses. This is quite simple, but in some cases one cannot afford to wait so long -- there might be other high-priority work to be done.}(hj& hj$ hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM hj hhubh)}(hdIn such cases, one uses call_rcu() rather than synchronize_rcu(). The call_rcu() API is as follows::h]hcIn such cases, one uses call_rcu() rather than synchronize_rcu(). The call_rcu() API is as follows:}(hcIn such cases, one uses call_rcu() rather than synchronize_rcu(). The call_rcu() API is as follows:hj2 hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM hj hhubj)}(h:void call_rcu(struct rcu_head *head, rcu_callback_t func);h]h:void call_rcu(struct rcu_head *head, rcu_callback_t func);}(hhhjA ubah}(h]h ]h"]h$]h&]jjuh1jhhhMhj hhubh)}(hXThis function invokes func(head) after a grace period has elapsed. This invocation might happen from either softirq or process context, so the function is not permitted to block. The foo struct needs to have an rcu_head structure added, perhaps as follows::h]hXThis function invokes func(head) after a grace period has elapsed. This invocation might happen from either softirq or process context, so the function is not permitted to block. The foo struct needs to have an rcu_head structure added, perhaps as follows:}(hXThis function invokes func(head) after a grace period has elapsed. This invocation might happen from either softirq or process context, so the function is not permitted to block. The foo struct needs to have an rcu_head structure added, perhaps as follows:hjO hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubj)}(h[struct foo { int a; char b; long c; struct rcu_head rcu; };h]h[struct foo { int a; char b; long c; struct rcu_head rcu; };}(hhhj^ ubah}(h]h ]h"]h$]h&]jjuh1jhhhMhj hhubh)}(h>The foo_update_a() function might then be written as follows::h]h=The foo_update_a() function might then be written as follows:}(h=The foo_update_a() function might then be written as follows:hjl hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubj)}(hX/* * Create a new struct foo that is the same as the one currently * pointed to by gbl_foo, except that field "a" is replaced * with "new_a". Points gbl_foo to the new structure, and * frees up the old structure after a grace period. * * Uses rcu_assign_pointer() to ensure that concurrent readers * see the initialized version of the new structure. * * Uses call_rcu() to ensure that any readers that might have * references to the old structure complete before freeing the * old structure. */ void foo_update_a(int new_a) { struct foo *new_fp; struct foo *old_fp; new_fp = kmalloc(sizeof(*new_fp), GFP_KERNEL); spin_lock(&foo_mutex); old_fp = rcu_dereference_protected(gbl_foo, lockdep_is_held(&foo_mutex)); *new_fp = *old_fp; new_fp->a = new_a; rcu_assign_pointer(gbl_foo, new_fp); spin_unlock(&foo_mutex); call_rcu(&old_fp->rcu, foo_reclaim); }h]hX/* * Create a new struct foo that is the same as the one currently * pointed to by gbl_foo, except that field "a" is replaced * with "new_a". Points gbl_foo to the new structure, and * frees up the old structure after a grace period. * * Uses rcu_assign_pointer() to ensure that concurrent readers * see the initialized version of the new structure. * * Uses call_rcu() to ensure that any readers that might have * references to the old structure complete before freeing the * old structure. */ void foo_update_a(int new_a) { struct foo *new_fp; struct foo *old_fp; new_fp = kmalloc(sizeof(*new_fp), GFP_KERNEL); spin_lock(&foo_mutex); old_fp = rcu_dereference_protected(gbl_foo, lockdep_is_held(&foo_mutex)); *new_fp = *old_fp; new_fp->a = new_a; rcu_assign_pointer(gbl_foo, new_fp); spin_unlock(&foo_mutex); call_rcu(&old_fp->rcu, foo_reclaim); }}(hhhj{ ubah}(h]h ]h"]h$]h&]jjuh1jhhhM hj hhubh)}(h4The foo_reclaim() function might appear as follows::h]h3The foo_reclaim() function might appear as follows:}(h3The foo_reclaim() function might appear as follows:hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM<hj hhubj)}(hvoid foo_reclaim(struct rcu_head *rp) { struct foo *fp = container_of(rp, struct foo, rcu); foo_cleanup(fp->a); kfree(fp); }h]hvoid foo_reclaim(struct rcu_head *rp) { struct foo *fp = container_of(rp, struct foo, rcu); foo_cleanup(fp->a); kfree(fp); }}(hhhj ubah}(h]h ]h"]h$]h&]jjuh1jhhhM>hj hhubh)}(hThe container_of() primitive is a macro that, given a pointer into a struct, the type of the struct, and the pointed-to field within the struct, returns a pointer to the beginning of the struct.h]hThe container_of() primitive is a macro that, given a pointer into a struct, the type of the struct, and the pointed-to field within the struct, returns a pointer to the beginning of the struct.}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMGhj hhubh)}(hX&The use of call_rcu() permits the caller of foo_update_a() to immediately regain control, without needing to worry further about the old version of the newly updated element. It also clearly shows the RCU distinction between updater, namely foo_update_a(), and reclaimer, namely foo_reclaim().h]hX&The use of call_rcu() permits the caller of foo_update_a() to immediately regain control, without needing to worry further about the old version of the newly updated element. It also clearly shows the RCU distinction between updater, namely foo_update_a(), and reclaimer, namely foo_reclaim().}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMKhj hhubh)}(hThe summary of advice is the same as for the previous section, except that we are now using call_rcu() rather than synchronize_rcu():h]hThe summary of advice is the same as for the previous section, except that we are now using call_rcu() rather than synchronize_rcu():}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMQhj hhubj0 )}(hhh]jZ)}(hUse call_rcu() **after** removing a data element from an RCU-protected data structure in order to register a callback function that will be invoked after the completion of all RCU read-side critical sections that might be referencing that data item. h]h)}(hUse call_rcu() **after** removing a data element from an RCU-protected data structure in order to register a callback function that will be invoked after the completion of all RCU read-side critical sections that might be referencing that data item.h](hUse call_rcu() }(hUse call_rcu() hj hhhNhNubj)}(h **after**h]hafter}(hhhj hhhNhNubah}(h]h ]h"]h$]h&]uh1jhj ubh removing a data element from an RCU-protected data structure in order to register a callback function that will be invoked after the completion of all RCU read-side critical sections that might be referencing that data item.}(h removing a data element from an RCU-protected data structure in order to register a callback function that will be invoked after the completion of all RCU read-side critical sections that might be referencing that data item.hj hhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMThj ubah}(h]h ]h"]h$]h&]uh1jYhj hhhhhNubah}(h]h ]h"]h$]h&]j j uh1j/ hhhMThj hhubh)}(hIf the callback for call_rcu() is not doing anything more than calling kfree() on the structure, you can use kfree_rcu() instead of call_rcu() to avoid having to write your own callback::h]hIf the callback for call_rcu() is not doing anything more than calling kfree() on the structure, you can use kfree_rcu() instead of call_rcu() to avoid having to write your own callback:}(hIf the callback for call_rcu() is not doing anything more than calling kfree() on the structure, you can use kfree_rcu() instead of call_rcu() to avoid having to write your own callback:hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMZhj hhubj)}(hkfree_rcu(old_fp, rcu);h]hkfree_rcu(old_fp, rcu);}(hhhj ubah}(h]h ]h"]h$]h&]jjuh1jhhhM^hj hhubh)}(h|If the occasional sleep is permitted, the single-argument form may be used, omitting the rcu_head structure from struct foo.h]h|If the occasional sleep is permitted, the single-argument form may be used, omitting the rcu_head structure from struct foo.}(hj$ hj" hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM`hj hhubj)}(hhh]h)}(hkfree_rcu_mightsleep(old_fp);h]hkfree_rcu_mightsleep(old_fp);}(hj5 hj3 hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMchj0 ubah}(h]h ]h"]h$]h&]uh1jhj hhhhhNubh)}(hyThis variant almost never blocks, but might do so by invoking synchronize_rcu() in response to memory-allocation failure.h]hyThis variant almost never blocks, but might do so by invoking synchronize_rcu() in response to memory-allocation failure.}(hjI hjG hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMehj hhubh)}(hGAgain, see checklist.rst for additional rules governing the use of RCU.h]hGAgain, see checklist.rst for additional rules governing the use of RCU.}(hjW hjU hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhhj hhubh)}(h.. _5_whatisRCU:h]h}(h]h ]h"]h$]h&]hid8uh1hhMjhj hhhhubeh}(h]('what-if-my-updating-thread-cannot-blockj eh ]h"](+4. what if my updating thread cannot block? 4_whatisrcueh$]h&]uh1hhhhhhhhMj}jt j sj}j j subh)}(hhh](h)}(h05. WHAT ARE SOME SIMPLE IMPLEMENTATIONS OF RCU?h]h05. WHAT ARE SOME SIMPLE IMPLEMENTATIONS OF RCU?}(hj~ hj| hhhNhNubah}(h]h ]h"]h$]h&]uh1hhjy hhhhhMmubh)}(hX\One of the nice things about RCU is that it has extremely simple "toy" implementations that are a good first step towards understanding the production-quality implementations in the Linux kernel. This section presents two such "toy" implementations of RCU, one that is implemented in terms of familiar locking primitives, and another that more closely resembles "classic" RCU. Both are way too simple for real-world use, lacking both functionality and performance. However, they are useful in getting a feel for how RCU works. See kernel/rcu/update.c for a production-quality implementation, and see:h]hXhOne of the nice things about RCU is that it has extremely simple “toy” implementations that are a good first step towards understanding the production-quality implementations in the Linux kernel. This section presents two such “toy” implementations of RCU, one that is implemented in terms of familiar locking primitives, and another that more closely resembles “classic” RCU. Both are way too simple for real-world use, lacking both functionality and performance. However, they are useful in getting a feel for how RCU works. See kernel/rcu/update.c for a production-quality implementation, and see:}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMohjy hhubj)}(hhh]h)}(hThttps://docs.google.com/document/d/1X0lThx8OK0ZgLMqVoXiR4ZrGURHrXK6NyLRbeXe3Xac/edith]h)}(hj h]hThttps://docs.google.com/document/d/1X0lThx8OK0ZgLMqVoXiR4ZrGURHrXK6NyLRbeXe3Xac/edit}(hhhj hhhNhNubah}(h]h ]h"]h$]h&]refurij uh1hhj ubah}(h]h ]h"]h$]h&]uh1hhhhMyhj ubah}(h]h ]h"]h$]h&]uh1jhjy hhhhhNubh)}(hfor papers describing the Linux kernel RCU implementation. The OLS'01 and OLS'02 papers are a good introduction, and the dissertation provides more details on the current implementation as of early 2004.h]hfor papers describing the Linux kernel RCU implementation. The OLS’01 and OLS’02 papers are a good introduction, and the dissertation provides more details on the current implementation as of early 2004.}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM{hjy hhubh)}(hhh](h)}(h%5A. "TOY" IMPLEMENTATION #1: LOCKINGh]h)5A. “TOY” IMPLEMENTATION #1: LOCKING}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhj hhhhhMubh)}(hXThis section presents a "toy" RCU implementation that is based on familiar locking primitives. Its overhead makes it a non-starter for real-life use, as does its lack of scalability. It is also unsuitable for realtime use, since it allows scheduling latency to "bleed" from one read-side critical section to another. It also assumes recursive reader-writer locks: If you try this with non-recursive locks, and you allow nested rcu_read_lock() calls, you can deadlock.h]hXThis section presents a “toy” RCU implementation that is based on familiar locking primitives. Its overhead makes it a non-starter for real-life use, as does its lack of scalability. It is also unsuitable for realtime use, since it allows scheduling latency to “bleed” from one read-side critical section to another. It also assumes recursive reader-writer locks: If you try this with non-recursive locks, and you allow nested rcu_read_lock() calls, you can deadlock.}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(h]However, it is probably the easiest implementation to relate to, so is a good starting point.h]h]However, it is probably the easiest implementation to relate to, so is a good starting point.}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(hIt is extremely simple::h]hIt is extremely simple:}(hIt is extremely simple:hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubj)}(hX1static DEFINE_RWLOCK(rcu_gp_mutex); void rcu_read_lock(void) { read_lock(&rcu_gp_mutex); } void rcu_read_unlock(void) { read_unlock(&rcu_gp_mutex); } void synchronize_rcu(void) { write_lock(&rcu_gp_mutex); smp_mb__after_spinlock(); write_unlock(&rcu_gp_mutex); }h]hX1static DEFINE_RWLOCK(rcu_gp_mutex); void rcu_read_lock(void) { read_lock(&rcu_gp_mutex); } void rcu_read_unlock(void) { read_unlock(&rcu_gp_mutex); } void synchronize_rcu(void) { write_lock(&rcu_gp_mutex); smp_mb__after_spinlock(); write_unlock(&rcu_gp_mutex); }}(hhhj ubah}(h]h ]h"]h$]h&]jjuh1jhhhMhj hhubh)}(h[You can ignore rcu_assign_pointer() and rcu_dereference() without missing much. But here are simplified versions anyway. And whatever you do, don't forget about them when submitting patches making use of RCU!]::h]h[You can ignore rcu_assign_pointer() and rcu_dereference() without missing much. But here are simplified versions anyway. And whatever you do, don’t forget about them when submitting patches making use of RCU!]:}(h[You can ignore rcu_assign_pointer() and rcu_dereference() without missing much. But here are simplified versions anyway. And whatever you do, don't forget about them when submitting patches making use of RCU!]:hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubj)}(h#define rcu_assign_pointer(p, v) \ ({ \ smp_store_release(&(p), (v)); \ }) #define rcu_dereference(p) \ ({ \ typeof(p) _________p1 = READ_ONCE(p); \ (_________p1); \ })h]h#define rcu_assign_pointer(p, v) \ ({ \ smp_store_release(&(p), (v)); \ }) #define rcu_dereference(p) \ ({ \ typeof(p) _________p1 = READ_ONCE(p); \ (_________p1); \ })}(hhhj ubah}(h]h ]h"]h$]h&]jjuh1jhhhMhj hhubh)}(hX]The rcu_read_lock() and rcu_read_unlock() primitive read-acquire and release a global reader-writer lock. The synchronize_rcu() primitive write-acquires this same lock, then releases it. This means that once synchronize_rcu() exits, all RCU read-side critical sections that were in progress before synchronize_rcu() was called are guaranteed to have completed -- there is no way that synchronize_rcu() would have been able to write-acquire the lock otherwise. The smp_mb__after_spinlock() promotes synchronize_rcu() to a full memory barrier in compliance with the "Memory-Barrier Guarantees" listed in:h]hXaThe rcu_read_lock() and rcu_read_unlock() primitive read-acquire and release a global reader-writer lock. The synchronize_rcu() primitive write-acquires this same lock, then releases it. This means that once synchronize_rcu() exits, all RCU read-side critical sections that were in progress before synchronize_rcu() was called are guaranteed to have completed -- there is no way that synchronize_rcu() would have been able to write-acquire the lock otherwise. The smp_mb__after_spinlock() promotes synchronize_rcu() to a full memory barrier in compliance with the “Memory-Barrier Guarantees” listed in:}(hj0 hj. hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubj)}(hhh]h)}(h$Design/Requirements/Requirements.rsth]h$Design/Requirements/Requirements.rst}(hjA hj? hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj< ubah}(h]h ]h"]h$]h&]uh1jhj hhhhhNubh)}(hXIt is possible to nest rcu_read_lock(), since reader-writer locks may be recursively acquired. Note also that rcu_read_lock() is immune from deadlock (an important property of RCU). The reason for this is that the only thing that can block rcu_read_lock() is a synchronize_rcu(). But synchronize_rcu() does not acquire any locks while holding rcu_gp_mutex, so there can be no deadlock cycle.h]hXIt is possible to nest rcu_read_lock(), since reader-writer locks may be recursively acquired. Note also that rcu_read_lock() is immune from deadlock (an important property of RCU). The reason for this is that the only thing that can block rcu_read_lock() is a synchronize_rcu(). But synchronize_rcu() does not acquire any locks while holding rcu_gp_mutex, so there can be no deadlock cycle.}(hjU hjS hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(h .. _quiz_1:h]h}(h]h ]h"]h$]h&]hquiz-1uh1hhMhj hhhhubhdefinition_list)}(hhh]hdefinition_list_item)}(hQuick Quiz #1: Why is this argument naive? How could a deadlock occur when using this algorithm in a real-world Linux kernel? How could this deadlock be avoided? h](hterm)}(hQuick Quiz #1:h]hQuick Quiz #1:}(hj{ hjy hhhNhNubah}(h]h ]h"]h$]h&]uh1jw hhhMhjs ubh definition)}(hhh]h)}(hWhy is this argument naive? How could a deadlock occur when using this algorithm in a real-world Linux kernel? How could this deadlock be avoided?h]hWhy is this argument naive? How could a deadlock occur when using this algorithm in a real-world Linux kernel? How could this deadlock be avoided?}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj ubah}(h]h ]h"]h$]h&]uh1j hjs ubeh}(h]h ]h"]h$]h&]uh1jq hhhMhjn ubah}(h]jk ah ]h"]quiz_1ah$]h&]uh1jl hj hhhhhNj}j ja sj}jk ja subh)}(h*:ref:`Answers to Quick Quiz <9_whatisRCU>`h]h)}(hj h]j)}(hj h]hAnswers to Quick Quiz}(hhhj hhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhj ubah}(h]h ]h"]h$]h&]refdocj refdomainj reftyperef refexplicitrefwarnj 9_whatisrcuuh1hhhhMhj ubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubeh}(h]a-toy-implementation-1-lockingah ]h"]$5a. "toy" implementation #1: lockingah$]h&]uh1hhjy hhhhhMubh)}(hhh](h)}(h"5B. "TOY" EXAMPLE #2: CLASSIC RCUh]h&5B. “TOY” EXAMPLE #2: CLASSIC RCU}(hj hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhj hhhhhMubh)}(hXnThis section presents a "toy" RCU implementation that is based on "classic RCU". It is also short on performance (but only for updates) and on features such as hotplug CPU and the ability to run in CONFIG_PREEMPTION kernels. The definitions of rcu_dereference() and rcu_assign_pointer() are the same as those shown in the preceding section, so they are omitted. ::h]hXsThis section presents a “toy” RCU implementation that is based on “classic RCU”. It is also short on performance (but only for updates) and on features such as hotplug CPU and the ability to run in CONFIG_PREEMPTION kernels. The definitions of rcu_dereference() and rcu_assign_pointer() are the same as those shown in the preceding section, so they are omitted.}(hXkThis section presents a "toy" RCU implementation that is based on "classic RCU". It is also short on performance (but only for updates) and on features such as hotplug CPU and the ability to run in CONFIG_PREEMPTION kernels. The definitions of rcu_dereference() and rcu_assign_pointer() are the same as those shown in the preceding section, so they are omitted.hj hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubj)}(hvoid rcu_read_lock(void) { } void rcu_read_unlock(void) { } void synchronize_rcu(void) { int cpu; for_each_possible_cpu(cpu) run_on(cpu); }h]hvoid rcu_read_lock(void) { } void rcu_read_unlock(void) { } void synchronize_rcu(void) { int cpu; for_each_possible_cpu(cpu) run_on(cpu); }}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMhj hhubh)}(hX0Note that rcu_read_lock() and rcu_read_unlock() do absolutely nothing. This is the great strength of classic RCU in a non-preemptive kernel: read-side overhead is precisely zero, at least on non-Alpha CPUs. And there is absolutely no way that rcu_read_lock() can possibly participate in a deadlock cycle!h]hX0Note that rcu_read_lock() and rcu_read_unlock() do absolutely nothing. This is the great strength of classic RCU in a non-preemptive kernel: read-side overhead is precisely zero, at least on non-Alpha CPUs. And there is absolutely no way that rcu_read_lock() can possibly participate in a deadlock cycle!}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(hXThe implementation of synchronize_rcu() simply schedules itself on each CPU in turn. The run_on() primitive can be implemented straightforwardly in terms of the sched_setaffinity() primitive. Of course, a somewhat less "toy" implementation would restore the affinity upon completion rather than just leaving all tasks running on the last CPU, but when I said "toy", I meant **toy**!h](hXThe implementation of synchronize_rcu() simply schedules itself on each CPU in turn. The run_on() primitive can be implemented straightforwardly in terms of the sched_setaffinity() primitive. Of course, a somewhat less “toy” implementation would restore the affinity upon completion rather than just leaving all tasks running on the last CPU, but when I said “toy”, I meant }(hXxThe implementation of synchronize_rcu() simply schedules itself on each CPU in turn. The run_on() primitive can be implemented straightforwardly in terms of the sched_setaffinity() primitive. Of course, a somewhat less "toy" implementation would restore the affinity upon completion rather than just leaving all tasks running on the last CPU, but when I said "toy", I meant hjhhhNhNubj)}(h**toy**h]htoy}(hhhj%hhhNhNubah}(h]h ]h"]h$]h&]uh1jhjubh!}(h!hjhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(h+So how the heck is this supposed to work???h]h+So how the heck is this supposed to work???}(hj@hj>hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(hXWRemember that it is illegal to block while in an RCU read-side critical section. Therefore, if a given CPU executes a context switch, we know that it must have completed all preceding RCU read-side critical sections. Once **all** CPUs have executed a context switch, then **all** preceding RCU read-side critical sections will have completed.h](hRemember that it is illegal to block while in an RCU read-side critical section. Therefore, if a given CPU executes a context switch, we know that it must have completed all preceding RCU read-side critical sections. Once }(hRemember that it is illegal to block while in an RCU read-side critical section. Therefore, if a given CPU executes a context switch, we know that it must have completed all preceding RCU read-side critical sections. Once hjLhhhNhNubj)}(h**all**h]hall}(hhhjUhhhNhNubah}(h]h ]h"]h$]h&]uh1jhjLubh+ CPUs have executed a context switch, then }(h+ CPUs have executed a context switch, then hjLhhhNhNubj)}(h**all**h]hall}(hhhjhhhhNhNubah}(h]h ]h"]h$]h&]uh1jhjLubh? preceding RCU read-side critical sections will have completed.}(h? preceding RCU read-side critical sections will have completed.hjLhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(hXSo, suppose that we remove a data item from its structure and then invoke synchronize_rcu(). Once synchronize_rcu() returns, we are guaranteed that there are no RCU read-side critical sections holding a reference to that data item, so we can safely reclaim it.h]hXSo, suppose that we remove a data item from its structure and then invoke synchronize_rcu(). Once synchronize_rcu() returns, we are guaranteed that there are no RCU read-side critical sections holding a reference to that data item, so we can safely reclaim it.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(h .. _quiz_2:h]h}(h]h ]h"]h$]h&]hquiz-2uh1hhMhj hhhhubjm )}(hhh]jr )}(hWQuick Quiz #2: Give an example where Classic RCU's read-side overhead is **negative**. h](jx )}(hQuick Quiz #2:h]hQuick Quiz #2:}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1jw hhhMhjubj )}(hhh]h)}(hGGive an example where Classic RCU's read-side overhead is **negative**.h](h`h]h)}(hjh]j)}(hjh]hAnswers to Quick Quiz}(hhhjhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdocj refdomainjreftyperef refexplicitrefwarnj 9_whatisrcuuh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1hhhhMhj hhubh)}(h .. _quiz_3:h]h}(h]h ]h"]h$]h&]hquiz-3uh1hhMhj hhhhubjm )}(hhh]jr )}(hQuick Quiz #3: If it is illegal to block in an RCU read-side critical section, what the heck do you do in CONFIG_PREEMPT_RT, where normal spinlocks can block??? h](jx )}(hQuick Quiz #3:h]hQuick Quiz #3:}(hj%hj#hhhNhNubah}(h]h ]h"]h$]h&]uh1jw hhhM hjubj )}(hhh]h)}(hIf it is illegal to block in an RCU read-side critical section, what the heck do you do in CONFIG_PREEMPT_RT, where normal spinlocks can block???h]hIf it is illegal to block in an RCU read-side critical section, what the heck do you do in CONFIG_PREEMPT_RT, where normal spinlocks can block???}(hj6hj4hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj1ubah}(h]h ]h"]h$]h&]uh1j hjubeh}(h]h ]h"]h$]h&]uh1jq hhhM hjubah}(h]jah ]h"]quiz_3ah$]h&]uh1jl hj hhhhhNj}jRjsj}jjsubh)}(h*:ref:`Answers to Quick Quiz <9_whatisRCU>`h]h)}(hjYh]j)}(hjYh]hAnswers to Quick Quiz}(hhhj^hhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhj[ubah}(h]h ]h"]h$]h&]refdocj refdomainjhreftyperef refexplicitrefwarnj 9_whatisrcuuh1hhhhM hjWubah}(h]h ]h"]h$]h&]uh1hhhhM hj hhubh)}(h.. _6_whatisRCU:h]h}(h]h ]h"]h$]h&]hid9uh1hhMhj hhhhubeh}(h]b-toy-example-2-classic-rcuah ]h"]!5b. "toy" example #2: classic rcuah$]h&]uh1hhjy hhhhhMubeh}(h](+what-are-some-simple-implementations-of-rcujm eh ]h"](/5. what are some simple implementations of rcu? 5_whatisrcueh$]h&]uh1hhhhhhhhMmj}jjc sj}jm jc subh)}(hhh](h)}(h&6. ANALOGY WITH READER-WRITER LOCKINGh]h&6. ANALOGY WITH READER-WRITER LOCKING}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhjhhhhhMubh)}(hAlthough RCU can be used in many different ways, a very common use of RCU is analogous to reader-writer locking. The following unified diff shows how closely related RCU and reader-writer locking can be. ::h]hAlthough RCU can be used in many different ways, a very common use of RCU is analogous to reader-writer locking. The following unified diff shows how closely related RCU and reader-writer locking can be.}(hAlthough RCU can be used in many different ways, a very common use of RCU is analogous to reader-writer locking. The following unified diff shows how closely related RCU and reader-writer locking can be.hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubj)}(hX@@ -5,5 +5,5 @@ struct el { int data; /* Other data fields */ }; -rwlock_t listmutex; +spinlock_t listmutex; struct el head; @@ -13,15 +14,15 @@ struct list_head *lp; struct el *p; - read_lock(&listmutex); - list_for_each_entry(p, head, lp) { + rcu_read_lock(); + list_for_each_entry_rcu(p, head, lp) { if (p->key == key) { *result = p->data; - read_unlock(&listmutex); + rcu_read_unlock(); return 1; } } - read_unlock(&listmutex); + rcu_read_unlock(); return 0; } @@ -29,15 +30,16 @@ { struct el *p; - write_lock(&listmutex); + spin_lock(&listmutex); list_for_each_entry(p, head, lp) { if (p->key == key) { - list_del(&p->list); - write_unlock(&listmutex); + list_del_rcu(&p->list); + spin_unlock(&listmutex); + synchronize_rcu(); kfree(p); return 1; } } - write_unlock(&listmutex); + spin_unlock(&listmutex); return 0; }h]hX@@ -5,5 +5,5 @@ struct el { int data; /* Other data fields */ }; -rwlock_t listmutex; +spinlock_t listmutex; struct el head; @@ -13,15 +14,15 @@ struct list_head *lp; struct el *p; - read_lock(&listmutex); - list_for_each_entry(p, head, lp) { + rcu_read_lock(); + list_for_each_entry_rcu(p, head, lp) { if (p->key == key) { *result = p->data; - read_unlock(&listmutex); + rcu_read_unlock(); return 1; } } - read_unlock(&listmutex); + rcu_read_unlock(); return 0; } @@ -29,15 +30,16 @@ { struct el *p; - write_lock(&listmutex); + spin_lock(&listmutex); list_for_each_entry(p, head, lp) { if (p->key == key) { - list_del(&p->list); - write_unlock(&listmutex); + list_del_rcu(&p->list); + spin_unlock(&listmutex); + synchronize_rcu(); kfree(p); return 1; } } - write_unlock(&listmutex); + spin_unlock(&listmutex); return 0; }}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMhjhhubh)}(h1Or, for those who prefer a side-by-side listing::h]h0Or, for those who prefer a side-by-side listing:}(h0Or, for those who prefer a side-by-side listing:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMJhjhhubj)}(hX1 struct el { 1 struct el { 2 struct list_head list; 2 struct list_head list; 3 long key; 3 long key; 4 spinlock_t mutex; 4 spinlock_t mutex; 5 int data; 5 int data; 6 /* Other data fields */ 6 /* Other data fields */ 7 }; 7 }; 8 rwlock_t listmutex; 8 spinlock_t listmutex; 9 struct el head; 9 struct el head;h]hX1 struct el { 1 struct el { 2 struct list_head list; 2 struct list_head list; 3 long key; 3 long key; 4 spinlock_t mutex; 4 spinlock_t mutex; 5 int data; 5 int data; 6 /* Other data fields */ 6 /* Other data fields */ 7 }; 7 }; 8 rwlock_t listmutex; 8 spinlock_t listmutex; 9 struct el head; 9 struct el head;}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMLhjhhubj)}(hX 1 int search(long key, int *result) 1 int search(long key, int *result) 2 { 2 { 3 struct list_head *lp; 3 struct list_head *lp; 4 struct el *p; 4 struct el *p; 5 5 6 read_lock(&listmutex); 6 rcu_read_lock(); 7 list_for_each_entry(p, head, lp) { 7 list_for_each_entry_rcu(p, head, lp) { 8 if (p->key == key) { 8 if (p->key == key) { 9 *result = p->data; 9 *result = p->data; 10 read_unlock(&listmutex); 10 rcu_read_unlock(); 11 return 1; 11 return 1; 12 } 12 } 13 } 13 } 14 read_unlock(&listmutex); 14 rcu_read_unlock(); 15 return 0; 15 return 0; 16 } 16 }h]hX 1 int search(long key, int *result) 1 int search(long key, int *result) 2 { 2 { 3 struct list_head *lp; 3 struct list_head *lp; 4 struct el *p; 4 struct el *p; 5 5 6 read_lock(&listmutex); 6 rcu_read_lock(); 7 list_for_each_entry(p, head, lp) { 7 list_for_each_entry_rcu(p, head, lp) { 8 if (p->key == key) { 8 if (p->key == key) { 9 *result = p->data; 9 *result = p->data; 10 read_unlock(&listmutex); 10 rcu_read_unlock(); 11 return 1; 11 return 1; 12 } 12 } 13 } 13 } 14 read_unlock(&listmutex); 14 rcu_read_unlock(); 15 return 0; 15 return 0; 16 } 16 }}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMXhjhhubj)}(hX 1 int delete(long key) 1 int delete(long key) 2 { 2 { 3 struct el *p; 3 struct el *p; 4 4 5 write_lock(&listmutex); 5 spin_lock(&listmutex); 6 list_for_each_entry(p, head, lp) { 6 list_for_each_entry(p, head, lp) { 7 if (p->key == key) { 7 if (p->key == key) { 8 list_del(&p->list); 8 list_del_rcu(&p->list); 9 write_unlock(&listmutex); 9 spin_unlock(&listmutex); 10 synchronize_rcu(); 10 kfree(p); 11 kfree(p); 11 return 1; 12 return 1; 12 } 13 } 13 } 14 } 14 write_unlock(&listmutex); 15 spin_unlock(&listmutex); 15 return 0; 16 return 0; 16 } 17 }h]hX 1 int delete(long key) 1 int delete(long key) 2 { 2 { 3 struct el *p; 3 struct el *p; 4 4 5 write_lock(&listmutex); 5 spin_lock(&listmutex); 6 list_for_each_entry(p, head, lp) { 6 list_for_each_entry(p, head, lp) { 7 if (p->key == key) { 7 if (p->key == key) { 8 list_del(&p->list); 8 list_del_rcu(&p->list); 9 write_unlock(&listmutex); 9 spin_unlock(&listmutex); 10 synchronize_rcu(); 10 kfree(p); 11 kfree(p); 11 return 1; 12 return 1; 12 } 13 } 13 } 14 } 14 write_unlock(&listmutex); 15 spin_unlock(&listmutex); 15 return 0; 16 return 0; 16 } 17 }}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMkhjhhubh)}(hEither way, the differences are quite small. Read-side locking moves to rcu_read_lock() and rcu_read_unlock, update-side locking moves from a reader-writer lock to a simple spinlock, and a synchronize_rcu() precedes the kfree().h]hEither way, the differences are quite small. Read-side locking moves to rcu_read_lock() and rcu_read_unlock, update-side locking moves from a reader-writer lock to a simple spinlock, and a synchronize_rcu() precedes the kfree().}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM}hjhhubh)}(hXVHowever, there is one potential catch: the read-side and update-side critical sections can now run concurrently. In many cases, this will not be a problem, but it is necessary to check carefully regardless. For example, if multiple independent list updates must be seen as a single atomic update, converting to RCU will require special care.h]hXVHowever, there is one potential catch: the read-side and update-side critical sections can now run concurrently. In many cases, this will not be a problem, but it is necessary to check carefully regardless. For example, if multiple independent list updates must be seen as a single atomic update, converting to RCU will require special care.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubh)}(hAlso, the presence of synchronize_rcu() means that the RCU version of delete() can now block. If this is a problem, there is a callback-based mechanism that never blocks, namely call_rcu() or kfree_rcu(), that can be used in place of synchronize_rcu().h]hAlso, the presence of synchronize_rcu() means that the RCU version of delete() can now block. If this is a problem, there is a callback-based mechanism that never blocks, namely call_rcu() or kfree_rcu(), that can be used in place of synchronize_rcu().}(hj#hj!hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubh)}(h.. _7_whatisRCU:h]h}(h]h ]h"]h$]h&]hid10uh1hhMhjhhhhubeh}(h]("analogy-with-reader-writer-lockingjeh ]h"](%6. analogy with reader-writer locking 6_whatisrcueh$]h&]uh1hhhhhhhhMj}j@jsj}jjsubh)}(hhh](h)}(h#7. ANALOGY WITH REFERENCE COUNTINGh]h#7. ANALOGY WITH REFERENCE COUNTING}(hjJhjHhhhNhNubah}(h]h ]h"]h$]h&]uh1hhjEhhhhhMubh)}(hThe reader-writer analogy (illustrated by the previous section) is not always the best way to think about using RCU. Another helpful analogy considers RCU an effective reference count on everything which is protected by RCU.h]hThe reader-writer analogy (illustrated by the previous section) is not always the best way to think about using RCU. Another helpful analogy considers RCU an effective reference count on everything which is protected by RCU.}(hjXhjVhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjEhhubh)}(hX'A reference count typically does not prevent the referenced object's values from changing, but does prevent changes to type -- particularly the gross change of type that happens when that object's memory is freed and re-allocated for some other purpose. Once a type-safe reference to the object is obtained, some other mechanism is needed to ensure consistent access to the data in the object. This could involve taking a spinlock, but with RCU the typical approach is to perform reads with SMP-aware operations such as smp_load_acquire(), to perform updates with atomic read-modify-write operations, and to provide the necessary ordering. RCU provides a number of support functions that embed the required operations and ordering, such as the list_for_each_entry_rcu() macro used in the previous section.h]hX+A reference count typically does not prevent the referenced object’s values from changing, but does prevent changes to type -- particularly the gross change of type that happens when that object’s memory is freed and re-allocated for some other purpose. Once a type-safe reference to the object is obtained, some other mechanism is needed to ensure consistent access to the data in the object. This could involve taking a spinlock, but with RCU the typical approach is to perform reads with SMP-aware operations such as smp_load_acquire(), to perform updates with atomic read-modify-write operations, and to provide the necessary ordering. RCU provides a number of support functions that embed the required operations and ordering, such as the list_for_each_entry_rcu() macro used in the previous section.}(hjfhjdhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjEhhubh)}(hXCA more focused view of the reference counting behavior is that, between rcu_read_lock() and rcu_read_unlock(), any reference taken with rcu_dereference() on a pointer marked as ``__rcu`` can be treated as though a reference-count on that object has been temporarily increased. This prevents the object from changing type. Exactly what this means will depend on normal expectations of objects of that type, but it typically includes that spinlocks can still be safely locked, normal reference counters can be safely manipulated, and ``__rcu`` pointers can be safely dereferenced.h](hA more focused view of the reference counting behavior is that, between rcu_read_lock() and rcu_read_unlock(), any reference taken with rcu_dereference() on a pointer marked as }(hA more focused view of the reference counting behavior is that, between rcu_read_lock() and rcu_read_unlock(), any reference taken with rcu_dereference() on a pointer marked as hjrhhhNhNubhliteral)}(h ``__rcu``h]h__rcu}(hhhj}hhhNhNubah}(h]h ]h"]h$]h&]uh1j{hjrubhX[ can be treated as though a reference-count on that object has been temporarily increased. This prevents the object from changing type. Exactly what this means will depend on normal expectations of objects of that type, but it typically includes that spinlocks can still be safely locked, normal reference counters can be safely manipulated, and }(hX[ can be treated as though a reference-count on that object has been temporarily increased. This prevents the object from changing type. Exactly what this means will depend on normal expectations of objects of that type, but it typically includes that spinlocks can still be safely locked, normal reference counters can be safely manipulated, and hjrhhhNhNubj|)}(h ``__rcu``h]h__rcu}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]uh1j{hjrubh% pointers can be safely dereferenced.}(h% pointers can be safely dereferenced.hjrhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMhjEhhubh)}(heSome operations that one might expect to see on an object for which an RCU reference is held include:h]heSome operations that one might expect to see on an object for which an RCU reference is held include:}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjEhhubj)}(hhh]j0 )}(hhh](jZ)}(hFCopying out data that is guaranteed to be stable by the object's type.h]h)}(hjh]hHCopying out data that is guaranteed to be stable by the object’s type.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjubjZ)}(haUsing kref_get_unless_zero() or similar to get a longer-term reference. This may fail of course.h]h)}(haUsing kref_get_unless_zero() or similar to get a longer-term reference. This may fail of course.h]haUsing kref_get_unless_zero() or similar to get a longer-term reference. This may fail of course.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjubjZ)}(hAcquiring a spinlock in the object, and checking if the object still is the expected object and if so, manipulating it freely. h]h)}(h~Acquiring a spinlock in the object, and checking if the object still is the expected object and if so, manipulating it freely.h]h~Acquiring a spinlock in the object, and checking if the object still is the expected object and if so, manipulating it freely.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjubeh}(h]h ]h"]h$]h&]j j uh1j/ hhhMhjubah}(h]h ]h"]h$]h&]uh1jhjEhhhNhNubh)}(hXThe understanding that RCU provides a reference that only prevents a change of type is particularly visible with objects allocated from a slab cache marked ``SLAB_TYPESAFE_BY_RCU``. RCU operations may yield a reference to an object from such a cache that has been concurrently freed and the memory reallocated to a completely different object, though of the same type. In this case RCU doesn't even protect the identity of the object from changing, only its type. So the object found may not be the one expected, but it will be one where it is safe to take a reference (and then potentially acquiring a spinlock), allowing subsequent code to check whether the identity matches expectations. It is tempting to simply acquire the spinlock without first taking the reference, but unfortunately any spinlock in a ``SLAB_TYPESAFE_BY_RCU`` object must be initialized after each and every call to kmem_cache_alloc(), which renders reference-free spinlock acquisition completely unsafe. Therefore, when using ``SLAB_TYPESAFE_BY_RCU``, make proper use of a reference counter. (Those willing to initialize their locks in a kmem_cache constructor may also use locking, including cache-friendly sequence locking.)h](hThe understanding that RCU provides a reference that only prevents a change of type is particularly visible with objects allocated from a slab cache marked }(hThe understanding that RCU provides a reference that only prevents a change of type is particularly visible with objects allocated from a slab cache marked hjhhhNhNubj|)}(h``SLAB_TYPESAFE_BY_RCU``h]hSLAB_TYPESAFE_BY_RCU}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]uh1j{hjubhX{. RCU operations may yield a reference to an object from such a cache that has been concurrently freed and the memory reallocated to a completely different object, though of the same type. In this case RCU doesn’t even protect the identity of the object from changing, only its type. So the object found may not be the one expected, but it will be one where it is safe to take a reference (and then potentially acquiring a spinlock), allowing subsequent code to check whether the identity matches expectations. It is tempting to simply acquire the spinlock without first taking the reference, but unfortunately any spinlock in a }(hXy. RCU operations may yield a reference to an object from such a cache that has been concurrently freed and the memory reallocated to a completely different object, though of the same type. In this case RCU doesn't even protect the identity of the object from changing, only its type. So the object found may not be the one expected, but it will be one where it is safe to take a reference (and then potentially acquiring a spinlock), allowing subsequent code to check whether the identity matches expectations. It is tempting to simply acquire the spinlock without first taking the reference, but unfortunately any spinlock in a hjhhhNhNubj|)}(h``SLAB_TYPESAFE_BY_RCU``h]hSLAB_TYPESAFE_BY_RCU}(hhhj,hhhNhNubah}(h]h ]h"]h$]h&]uh1j{hjubh object must be initialized after each and every call to kmem_cache_alloc(), which renders reference-free spinlock acquisition completely unsafe. Therefore, when using }(h object must be initialized after each and every call to kmem_cache_alloc(), which renders reference-free spinlock acquisition completely unsafe. Therefore, when using hjhhhNhNubj|)}(h``SLAB_TYPESAFE_BY_RCU``h]hSLAB_TYPESAFE_BY_RCU}(hhhj?hhhNhNubah}(h]h ]h"]h$]h&]uh1j{hjubh, make proper use of a reference counter. (Those willing to initialize their locks in a kmem_cache constructor may also use locking, including cache-friendly sequence locking.)}(h, make proper use of a reference counter. (Those willing to initialize their locks in a kmem_cache constructor may also use locking, including cache-friendly sequence locking.)hjhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMhjEhhubh)}(hXqWith traditional reference counting -- such as that implemented by the kref library in Linux -- there is typically code that runs when the last reference to an object is dropped. With kref, this is the function passed to kref_put(). When RCU is being used, such finalization code must not be run until all ``__rcu`` pointers referencing the object have been updated, and then a grace period has passed. Every remaining globally visible pointer to the object must be considered to be a potential counted reference, and the finalization code is typically run using call_rcu() only after all those pointers have been changed.h](hX4With traditional reference counting -- such as that implemented by the kref library in Linux -- there is typically code that runs when the last reference to an object is dropped. With kref, this is the function passed to kref_put(). When RCU is being used, such finalization code must not be run until all }(hX4With traditional reference counting -- such as that implemented by the kref library in Linux -- there is typically code that runs when the last reference to an object is dropped. With kref, this is the function passed to kref_put(). When RCU is being used, such finalization code must not be run until all hjXhhhNhNubj|)}(h ``__rcu``h]h__rcu}(hhhjahhhNhNubah}(h]h ]h"]h$]h&]uh1j{hjXubhX4 pointers referencing the object have been updated, and then a grace period has passed. Every remaining globally visible pointer to the object must be considered to be a potential counted reference, and the finalization code is typically run using call_rcu() only after all those pointers have been changed.}(hX4 pointers referencing the object have been updated, and then a grace period has passed. Every remaining globally visible pointer to the object must be considered to be a potential counted reference, and the finalization code is typically run using call_rcu() only after all those pointers have been changed.hjXhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMhjEhhubh)}(hXTo see how to choose between these two analogies -- of RCU as a reader-writer lock and RCU as a reference counting system -- it is useful to reflect on the scale of the thing being protected. The reader-writer lock analogy looks at larger multi-part objects such as a linked list and shows how RCU can facilitate concurrency while elements are added to, and removed from, the list. The reference-count analogy looks at the individual objects and looks at how they can be accessed safely within whatever whole they are a part of.h]hXTo see how to choose between these two analogies -- of RCU as a reader-writer lock and RCU as a reference counting system -- it is useful to reflect on the scale of the thing being protected. The reader-writer lock analogy looks at larger multi-part objects such as a linked list and shows how RCU can facilitate concurrency while elements are added to, and removed from, the list. The reference-count analogy looks at the individual objects and looks at how they can be accessed safely within whatever whole they are a part of.}(hj|hjzhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjEhhubh)}(h.. _8_whatisRCU:h]h}(h]h ]h"]h$]h&]hid11uh1hhMhjEhhhhubeh}(h](analogy-with-reference-countingj9eh ]h"]("7. analogy with reference counting 7_whatisrcueh$]h&]uh1hhhhhhhhMj}jj/sj}j9j/subh)}(hhh](h)}(h8. FULL LIST OF RCU APIsh]h8. FULL LIST OF RCU APIs}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhjhhhhhMubh)}(hThe RCU APIs are documented in docbook-format header comments in the Linux-kernel source code, but it helps to have a full list of the APIs, since there does not appear to be a way to categorize them in docbook. Here is the list, by category.h]hThe RCU APIs are documented in docbook-format header comments in the Linux-kernel source code, but it helps to have a full list of the APIs, since there does not appear to be a way to categorize them in docbook. Here is the list, by category.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubh)}(hRCU list traversal::h]hRCU list traversal:}(hRCU list traversal:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubj)}(hXlist_entry_rcu list_entry_lockless list_first_entry_rcu list_next_rcu list_for_each_entry_rcu list_for_each_entry_continue_rcu list_for_each_entry_from_rcu list_first_or_null_rcu list_next_or_null_rcu hlist_first_rcu hlist_next_rcu hlist_pprev_rcu hlist_for_each_entry_rcu hlist_for_each_entry_rcu_bh hlist_for_each_entry_from_rcu hlist_for_each_entry_continue_rcu hlist_for_each_entry_continue_rcu_bh hlist_nulls_first_rcu hlist_nulls_for_each_entry_rcu hlist_bl_first_rcu hlist_bl_for_each_entry_rcuh]hXlist_entry_rcu list_entry_lockless list_first_entry_rcu list_next_rcu list_for_each_entry_rcu list_for_each_entry_continue_rcu list_for_each_entry_from_rcu list_first_or_null_rcu list_next_or_null_rcu hlist_first_rcu hlist_next_rcu hlist_pprev_rcu hlist_for_each_entry_rcu hlist_for_each_entry_rcu_bh hlist_for_each_entry_from_rcu hlist_for_each_entry_continue_rcu hlist_for_each_entry_continue_rcu_bh hlist_nulls_first_rcu hlist_nulls_for_each_entry_rcu hlist_bl_first_rcu hlist_bl_for_each_entry_rcu}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMhjhhubh)}(hRCU pointer/list update::h]hRCU pointer/list update:}(hRCU pointer/list update:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubj)}(hXrcu_assign_pointer list_add_rcu list_add_tail_rcu list_del_rcu list_replace_rcu hlist_add_behind_rcu hlist_add_before_rcu hlist_add_head_rcu hlist_add_tail_rcu hlist_del_rcu hlist_del_init_rcu hlist_replace_rcu list_splice_init_rcu list_splice_tail_init_rcu hlist_nulls_del_init_rcu hlist_nulls_del_rcu hlist_nulls_add_head_rcu hlist_bl_add_head_rcu hlist_bl_del_init_rcu hlist_bl_del_rcu hlist_bl_set_first_rcuh]hXrcu_assign_pointer list_add_rcu list_add_tail_rcu list_del_rcu list_replace_rcu hlist_add_behind_rcu hlist_add_before_rcu hlist_add_head_rcu hlist_add_tail_rcu hlist_del_rcu hlist_del_init_rcu hlist_replace_rcu list_splice_init_rcu list_splice_tail_init_rcu hlist_nulls_del_init_rcu hlist_nulls_del_rcu hlist_nulls_add_head_rcu hlist_bl_add_head_rcu hlist_bl_del_init_rcu hlist_bl_del_rcu hlist_bl_set_first_rcu}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMhjhhubh)}(hRCU::h]hRCU:}(hRCU:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubj)}(hX+Critical sections Grace period Barrier rcu_read_lock synchronize_net rcu_barrier rcu_read_unlock synchronize_rcu rcu_dereference synchronize_rcu_expedited rcu_read_lock_held call_rcu rcu_dereference_check kfree_rcu rcu_dereference_protectedh]hX+Critical sections Grace period Barrier rcu_read_lock synchronize_net rcu_barrier rcu_read_unlock synchronize_rcu rcu_dereference synchronize_rcu_expedited rcu_read_lock_held call_rcu rcu_dereference_check kfree_rcu rcu_dereference_protected}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMhjhhubh)}(hbh::h]hbh:}(hbh:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM!hjhhubj)}(hX;Critical sections Grace period Barrier rcu_read_lock_bh call_rcu rcu_barrier rcu_read_unlock_bh synchronize_rcu [local_bh_disable] synchronize_rcu_expedited [and friends] rcu_dereference_bh rcu_dereference_bh_check rcu_dereference_bh_protected rcu_read_lock_bh_heldh]hX;Critical sections Grace period Barrier rcu_read_lock_bh call_rcu rcu_barrier rcu_read_unlock_bh synchronize_rcu [local_bh_disable] synchronize_rcu_expedited [and friends] rcu_dereference_bh rcu_dereference_bh_check rcu_dereference_bh_protected rcu_read_lock_bh_held}(hhhj#ubah}(h]h ]h"]h$]h&]jjuh1jhhhM#hjhhubh)}(hsched::h]hsched:}(hsched:hj1hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM.hjhhubj)}(hXCritical sections Grace period Barrier rcu_read_lock_sched call_rcu rcu_barrier rcu_read_unlock_sched synchronize_rcu [preempt_disable] synchronize_rcu_expedited [and friends] rcu_read_lock_sched_notrace rcu_read_unlock_sched_notrace rcu_dereference_sched rcu_dereference_sched_check rcu_dereference_sched_protected rcu_read_lock_sched_heldh]hXCritical sections Grace period Barrier rcu_read_lock_sched call_rcu rcu_barrier rcu_read_unlock_sched synchronize_rcu [preempt_disable] synchronize_rcu_expedited [and friends] rcu_read_lock_sched_notrace rcu_read_unlock_sched_notrace rcu_dereference_sched rcu_dereference_sched_check rcu_dereference_sched_protected rcu_read_lock_sched_held}(hhhj@ubah}(h]h ]h"]h$]h&]jjuh1jhhhM0hjhhubh)}(h RCU-Tasks::h]h RCU-Tasks:}(h RCU-Tasks:hjNhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM>hjhhubj)}(hCritical sections Grace period Barrier N/A call_rcu_tasks rcu_barrier_tasks synchronize_rcu_tasksh]hCritical sections Grace period Barrier N/A call_rcu_tasks rcu_barrier_tasks synchronize_rcu_tasks}(hhhj]ubah}(h]h ]h"]h$]h&]jjuh1jhhhM@hjhhubh)}(hRCU-Tasks-Rude::h]hRCU-Tasks-Rude:}(hRCU-Tasks-Rude:hjkhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMFhjhhubj)}(hCritical sections Grace period Barrier N/A call_rcu_tasks_rude rcu_barrier_tasks_rude synchronize_rcu_tasks_rudeh]hCritical sections Grace period Barrier N/A call_rcu_tasks_rude rcu_barrier_tasks_rude synchronize_rcu_tasks_rude}(hhhjzubah}(h]h ]h"]h$]h&]jjuh1jhhhMHhjhhubh)}(hRCU-Tasks-Trace::h]hRCU-Tasks-Trace:}(hRCU-Tasks-Trace:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMNhjhhubj)}(hCritical sections Grace period Barrier rcu_read_lock_trace call_rcu_tasks_trace rcu_barrier_tasks_trace rcu_read_unlock_trace synchronize_rcu_tasks_traceh]hCritical sections Grace period Barrier rcu_read_lock_trace call_rcu_tasks_trace rcu_barrier_tasks_trace rcu_read_unlock_trace synchronize_rcu_tasks_trace}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMPhjhhubh)}(hSRCU::h]hSRCU:}(hSRCU:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMVhjhhubj)}(hCritical sections Grace period Barrier srcu_read_lock call_srcu srcu_barrier srcu_read_unlock synchronize_srcu srcu_dereference synchronize_srcu_expedited srcu_dereference_check srcu_read_lock_heldh]hCritical sections Grace period Barrier srcu_read_lock call_srcu srcu_barrier srcu_read_unlock synchronize_srcu srcu_dereference synchronize_srcu_expedited srcu_dereference_check srcu_read_lock_held}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMXhjhhubh)}(hSRCU: Initialization/cleanup::h]hSRCU: Initialization/cleanup:}(hSRCU: Initialization/cleanup:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM`hjhhubj)}(hCDEFINE_SRCU DEFINE_STATIC_SRCU init_srcu_struct cleanup_srcu_structh]hCDEFINE_SRCU DEFINE_STATIC_SRCU init_srcu_struct cleanup_srcu_struct}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMbhjhhubh)}(h'All: lockdep-checked RCU utility APIs::h]h&All: lockdep-checked RCU utility APIs:}(h&All: lockdep-checked RCU utility APIs:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMghjhhubj)}(h RCU_LOCKDEP_WARN rcu_sleep_checkh]h RCU_LOCKDEP_WARN rcu_sleep_check}(hhhjubah}(h]h ]h"]h$]h&]jjuh1jhhhMihjhhubh)}(h-All: Unchecked RCU-protected pointer access::h]h,All: Unchecked RCU-protected pointer access:}(h,All: Unchecked RCU-protected pointer access:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMlhjhhubj)}(hrcu_dereference_rawh]hrcu_dereference_raw}(hhhj ubah}(h]h ]h"]h$]h&]jjuh1jhhhMnhjhhubh)}(hKAll: Unchecked RCU-protected pointer access with dereferencing prohibited::h]hJAll: Unchecked RCU-protected pointer access with dereferencing prohibited:}(hJAll: Unchecked RCU-protected pointer access with dereferencing prohibited:hjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMphjhhubj)}(hrcu_access_pointerh]hrcu_access_pointer}(hhhj(ubah}(h]h ]h"]h$]h&]jjuh1jhhhMrhjhhubh)}(heSee the comment headers in the source code (or the docbook generated from them) for more information.h]heSee the comment headers in the source code (or the docbook generated from them) for more information.}(hj8hj6hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMthjhhubh)}(hHowever, given that there are no fewer than four families of RCU APIs in the Linux kernel, how do you choose which one to use? The following list can be helpful:h]hHowever, given that there are no fewer than four families of RCU APIs in the Linux kernel, how do you choose which one to use? The following list can be helpful:}(hjFhjDhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMwhjhhubjU)}(hhh](jZ)}(h3Will readers need to block? If so, you need SRCU. h]h)}(h2Will readers need to block? If so, you need SRCU.h]h2Will readers need to block? If so, you need SRCU.}(hj[hjYhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM{hjUubah}(h]h ]h"]h$]h&]uh1jYhjRhhhhhNubjZ)}(hWill readers need to block and are you doing tracing, for example, ftrace or BPF? If so, you need RCU-tasks, RCU-tasks-rude, and/or RCU-tasks-trace. h]h)}(hWill readers need to block and are you doing tracing, for example, ftrace or BPF? If so, you need RCU-tasks, RCU-tasks-rude, and/or RCU-tasks-trace.h]hWill readers need to block and are you doing tracing, for example, ftrace or BPF? If so, you need RCU-tasks, RCU-tasks-rude, and/or RCU-tasks-trace.}(hjshjqhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhM}hjmubah}(h]h ]h"]h$]h&]uh1jYhjRhhhhhNubjZ)}(hX#What about the -rt patchset? If readers would need to block in an non-rt kernel, you need SRCU. If readers would block when acquiring spinlocks in a -rt kernel, but not in a non-rt kernel, SRCU is not necessary. (The -rt patchset turns spinlocks into sleeplocks, hence this distinction.) h]h)}(hX"What about the -rt patchset? If readers would need to block in an non-rt kernel, you need SRCU. If readers would block when acquiring spinlocks in a -rt kernel, but not in a non-rt kernel, SRCU is not necessary. (The -rt patchset turns spinlocks into sleeplocks, hence this distinction.)h]hX"What about the -rt patchset? If readers would need to block in an non-rt kernel, you need SRCU. If readers would block when acquiring spinlocks in a -rt kernel, but not in a non-rt kernel, SRCU is not necessary. (The -rt patchset turns spinlocks into sleeplocks, hence this distinction.)}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjRhhhhhNubjZ)}(hXsDo you need to treat NMI handlers, hardirq handlers, and code segments with preemption disabled (whether via preempt_disable(), local_irq_save(), local_bh_disable(), or some other mechanism) as if they were explicit RCU readers? If so, RCU-sched readers are the only choice that will work for you, but since about v4.20 you use can use the vanilla RCU update primitives. h]h)}(hXrDo you need to treat NMI handlers, hardirq handlers, and code segments with preemption disabled (whether via preempt_disable(), local_irq_save(), local_bh_disable(), or some other mechanism) as if they were explicit RCU readers? If so, RCU-sched readers are the only choice that will work for you, but since about v4.20 you use can use the vanilla RCU update primitives.h]hXrDo you need to treat NMI handlers, hardirq handlers, and code segments with preemption disabled (whether via preempt_disable(), local_irq_save(), local_bh_disable(), or some other mechanism) as if they were explicit RCU readers? If so, RCU-sched readers are the only choice that will work for you, but since about v4.20 you use can use the vanilla RCU update primitives.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjRhhhhhNubjZ)}(hXgDo you need RCU grace periods to complete even in the face of softirq monopolization of one or more of the CPUs? For example, is your code subject to network-based denial-of-service attacks? If so, you should disable softirq across your readers, for example, by using rcu_read_lock_bh(). Since about v4.20 you use can use the vanilla RCU update primitives. h]h)}(hXfDo you need RCU grace periods to complete even in the face of softirq monopolization of one or more of the CPUs? For example, is your code subject to network-based denial-of-service attacks? If so, you should disable softirq across your readers, for example, by using rcu_read_lock_bh(). Since about v4.20 you use can use the vanilla RCU update primitives.h]hXfDo you need RCU grace periods to complete even in the face of softirq monopolization of one or more of the CPUs? For example, is your code subject to network-based denial-of-service attacks? If so, you should disable softirq across your readers, for example, by using rcu_read_lock_bh(). Since about v4.20 you use can use the vanilla RCU update primitives.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjRhhhhhNubjZ)}(hIs your workload too update-intensive for normal use of RCU, but inappropriate for other synchronization mechanisms? If so, consider SLAB_TYPESAFE_BY_RCU (which was originally named SLAB_DESTROY_BY_RCU). But please be careful! h]h)}(hIs your workload too update-intensive for normal use of RCU, but inappropriate for other synchronization mechanisms? If so, consider SLAB_TYPESAFE_BY_RCU (which was originally named SLAB_DESTROY_BY_RCU). But please be careful!h]hIs your workload too update-intensive for normal use of RCU, but inappropriate for other synchronization mechanisms? If so, consider SLAB_TYPESAFE_BY_RCU (which was originally named SLAB_DESTROY_BY_RCU). But please be careful!}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjRhhhhhNubjZ)}(hX7Do you need read-side critical sections that are respected even on CPUs that are deep in the idle loop, during entry to or exit from user-mode execution, or on an offlined CPU? If so, SRCU and RCU Tasks Trace are the only choices that will work for you, with SRCU being strongly preferred in almost all cases. h]h)}(hX6Do you need read-side critical sections that are respected even on CPUs that are deep in the idle loop, during entry to or exit from user-mode execution, or on an offlined CPU? If so, SRCU and RCU Tasks Trace are the only choices that will work for you, with SRCU being strongly preferred in almost all cases.h]hX6Do you need read-side critical sections that are respected even on CPUs that are deep in the idle loop, during entry to or exit from user-mode execution, or on an offlined CPU? If so, SRCU and RCU Tasks Trace are the only choices that will work for you, with SRCU being strongly preferred in almost all cases.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjRhhhhhNubjZ)}(hOtherwise, use RCU. h]h)}(hOtherwise, use RCU.h]hOtherwise, use RCU.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjRhhhhhNubeh}(h]h ]h"]h$]h&]jjjhjjuh1jThjhhhhhM{ubh)}(heOf course, this all assumes that you have determined that RCU is in fact the right tool for your job.h]heOf course, this all assumes that you have determined that RCU is in fact the right tool for your job.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjhhubh)}(h.. _9_whatisRCU:h]h}(h]h ]h"]h$]h&]hid12uh1hhMhjhhhhubeh}(h](full-list-of-rcu-apisjeh ]h"](8. full list of rcu apis 8_whatisrcueh$]h&]uh1hhhhhhhhMj}j:jsj}jjsubh)}(hhh](h)}(h9. ANSWERS TO QUICK QUIZZESh]h9. ANSWERS TO QUICK QUIZZES}(hjDhjBhhhNhNubah}(h]h ]h"]h$]h&]uh1hhj?hhhhhMubjm )}(hhh](jr )}(hQuick Quiz #1: Why is this argument naive? How could a deadlock occur when using this algorithm in a real-world Linux kernel? [Referring to the lock-based "toy" RCU algorithm.] h](jx )}(hQuick Quiz #1:h]hQuick Quiz #1:}(hjYhjWhhhNhNubah}(h]h ]h"]h$]h&]uh1jw hhhMhjSubj )}(hhh]h)}(hWhy is this argument naive? How could a deadlock occur when using this algorithm in a real-world Linux kernel? [Referring to the lock-based "toy" RCU algorithm.]h]hWhy is this argument naive? How could a deadlock occur when using this algorithm in a real-world Linux kernel? [Referring to the lock-based “toy” RCU algorithm.]}(hjjhjhhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjeubah}(h]h ]h"]h$]h&]uh1j hjSubeh}(h]h ]h"]h$]h&]uh1jq hhhMhjPubjr )}(hXAnswer: Consider the following sequence of events: 1. CPU 0 acquires some unrelated lock, call it "problematic_lock", disabling irq via spin_lock_irqsave(). 2. CPU 1 enters synchronize_rcu(), write-acquiring rcu_gp_mutex. 3. CPU 0 enters rcu_read_lock(), but must wait because CPU 1 holds rcu_gp_mutex. 4. CPU 1 is interrupted, and the irq handler attempts to acquire problematic_lock. The system is now deadlocked. One way to avoid this deadlock is to use an approach like that of CONFIG_PREEMPT_RT, where all normal spinlocks become blocking locks, and all irq handlers execute in the context of special tasks. In this case, in step 4 above, the irq handler would block, allowing CPU 1 to release rcu_gp_mutex, avoiding the deadlock. Even in the absence of deadlock, this RCU implementation allows latency to "bleed" from readers to other readers through synchronize_rcu(). To see this, consider task A in an RCU read-side critical section (thus read-holding rcu_gp_mutex), task B blocked attempting to write-acquire rcu_gp_mutex, and task C blocked in rcu_read_lock() attempting to read_acquire rcu_gp_mutex. Task A's RCU read-side latency is holding up task C, albeit indirectly via task B. Realtime RCU implementations therefore use a counter-based approach where tasks in RCU read-side critical sections cannot be blocked by tasks executing synchronize_rcu(). h](jx )}(hAnswer:h]hAnswer:}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1jw hhhMhjubj )}(hhh](h)}(h*Consider the following sequence of events:h]h*Consider the following sequence of events:}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubjU)}(hhh](jZ)}(hgCPU 0 acquires some unrelated lock, call it "problematic_lock", disabling irq via spin_lock_irqsave(). h]h)}(hfCPU 0 acquires some unrelated lock, call it "problematic_lock", disabling irq via spin_lock_irqsave().h]hjCPU 0 acquires some unrelated lock, call it “problematic_lock”, disabling irq via spin_lock_irqsave().}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjubjZ)}(h>CPU 1 enters synchronize_rcu(), write-acquiring rcu_gp_mutex. h]h)}(h=CPU 1 enters synchronize_rcu(), write-acquiring rcu_gp_mutex.h]h=CPU 1 enters synchronize_rcu(), write-acquiring rcu_gp_mutex.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjubjZ)}(hNCPU 0 enters rcu_read_lock(), but must wait because CPU 1 holds rcu_gp_mutex. h]h)}(hMCPU 0 enters rcu_read_lock(), but must wait because CPU 1 holds rcu_gp_mutex.h]hMCPU 0 enters rcu_read_lock(), but must wait because CPU 1 holds rcu_gp_mutex.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjubjZ)}(hPCPU 1 is interrupted, and the irq handler attempts to acquire problematic_lock. h]h)}(hOCPU 1 is interrupted, and the irq handler attempts to acquire problematic_lock.h]hOCPU 1 is interrupted, and the irq handler attempts to acquire problematic_lock.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1jYhjubeh}(h]h ]h"]h$]h&]jarabicjhjjuh1jThjubh)}(hThe system is now deadlocked.h]hThe system is now deadlocked.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubh)}(hX@One way to avoid this deadlock is to use an approach like that of CONFIG_PREEMPT_RT, where all normal spinlocks become blocking locks, and all irq handlers execute in the context of special tasks. In this case, in step 4 above, the irq handler would block, allowing CPU 1 to release rcu_gp_mutex, avoiding the deadlock.h]hX@One way to avoid this deadlock is to use an approach like that of CONFIG_PREEMPT_RT, where all normal spinlocks become blocking locks, and all irq handlers execute in the context of special tasks. In this case, in step 4 above, the irq handler would block, allowing CPU 1 to release rcu_gp_mutex, avoiding the deadlock.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubh)}(hXEven in the absence of deadlock, this RCU implementation allows latency to "bleed" from readers to other readers through synchronize_rcu(). To see this, consider task A in an RCU read-side critical section (thus read-holding rcu_gp_mutex), task B blocked attempting to write-acquire rcu_gp_mutex, and task C blocked in rcu_read_lock() attempting to read_acquire rcu_gp_mutex. Task A's RCU read-side latency is holding up task C, albeit indirectly via task B.h]hXEven in the absence of deadlock, this RCU implementation allows latency to “bleed” from readers to other readers through synchronize_rcu(). To see this, consider task A in an RCU read-side critical section (thus read-holding rcu_gp_mutex), task B blocked attempting to write-acquire rcu_gp_mutex, and task C blocked in rcu_read_lock() attempting to read_acquire rcu_gp_mutex. Task A’s RCU read-side latency is holding up task C, albeit indirectly via task B.}(hj-hj+hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubh)}(hRealtime RCU implementations therefore use a counter-based approach where tasks in RCU read-side critical sections cannot be blocked by tasks executing synchronize_rcu().h]hRealtime RCU implementations therefore use a counter-based approach where tasks in RCU read-side critical sections cannot be blocked by tasks executing synchronize_rcu().}(hj;hj9hhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubeh}(h]h ]h"]h$]h&]uh1j hjubeh}(h]h ]h"]h$]h&]uh1jq hhhMhjPhhubeh}(h]h ]h"]h$]h&]uh1jl hj?hhhhhNubh)}(h%:ref:`Back to Quick Quiz #1 `h]h)}(hj[h]j)}(hj[h]hBack to Quick Quiz #1}(hhhj`hhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhj]ubah}(h]h ]h"]h$]h&]refdocj refdomainjjreftyperef refexplicitrefwarnjquiz_1uh1hhhhMhjYubah}(h]h ]h"]h$]h&]uh1hhhhMhj?hhubjm )}(hhh](jr )}(hWQuick Quiz #2: Give an example where Classic RCU's read-side overhead is **negative**. h](jx )}(hQuick Quiz #2:h]hQuick Quiz #2:}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1jw hhhMhjubj )}(hhh]h)}(hGGive an example where Classic RCU's read-side overhead is **negative**.h](h`h]h)}(hjh]j)}(hjh]hBack to Quick Quiz #2}(hhhjhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdocj refdomainj)reftyperef refexplicitrefwarnjquiz_2uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1hhhhMhj?hhubjm )}(hhh](jr )}(hQuick Quiz #3: If it is illegal to block in an RCU read-side critical section, what the heck do you do in CONFIG_PREEMPT_RT, where normal spinlocks can block??? h](jx )}(hQuick Quiz #3:h]hQuick Quiz #3:}(hjJhjHhhhNhNubah}(h]h ]h"]h$]h&]uh1jw hhhMhjDubj )}(hhh]h)}(hIf it is illegal to block in an RCU read-side critical section, what the heck do you do in CONFIG_PREEMPT_RT, where normal spinlocks can block???h]hIf it is illegal to block in an RCU read-side critical section, what the heck do you do in CONFIG_PREEMPT_RT, where normal spinlocks can block???}(hj[hjYhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjVubah}(h]h ]h"]h$]h&]uh1j hjDubeh}(h]h ]h"]h$]h&]uh1jq hhhMhjAubjr )}(hX;Answer: Just as CONFIG_PREEMPT_RT permits preemption of spinlock critical sections, it permits preemption of RCU read-side critical sections. It also permits spinlocks blocking while in RCU read-side critical sections. Why the apparent inconsistency? Because it is possible to use priority boosting to keep the RCU grace periods short if need be (for example, if running short of memory). In contrast, if blocking waiting for (say) network reception, there is no way to know what should be boosted. Especially given that the process we need to boost might well be a human being who just went out for a pizza or something. And although a computer-operated cattle prod might arouse serious interest, it might also provoke serious objections. Besides, how does the computer know what pizza parlor the human being went to??? h](jx )}(hAnswer:h]hAnswer:}(hjyhjwhhhNhNubah}(h]h ]h"]h$]h&]uh1jw hhhMhjsubj )}(hhh](h)}(hJust as CONFIG_PREEMPT_RT permits preemption of spinlock critical sections, it permits preemption of RCU read-side critical sections. It also permits spinlocks blocking while in RCU read-side critical sections.h]hJust as CONFIG_PREEMPT_RT permits preemption of spinlock critical sections, it permits preemption of RCU read-side critical sections. It also permits spinlocks blocking while in RCU read-side critical sections.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubh)}(hX]Why the apparent inconsistency? Because it is possible to use priority boosting to keep the RCU grace periods short if need be (for example, if running short of memory). In contrast, if blocking waiting for (say) network reception, there is no way to know what should be boosted. Especially given that the process we need to boost might well be a human being who just went out for a pizza or something. And although a computer-operated cattle prod might arouse serious interest, it might also provoke serious objections. Besides, how does the computer know what pizza parlor the human being went to???h]hX]Why the apparent inconsistency? Because it is possible to use priority boosting to keep the RCU grace periods short if need be (for example, if running short of memory). In contrast, if blocking waiting for (say) network reception, there is no way to know what should be boosted. Especially given that the process we need to boost might well be a human being who just went out for a pizza or something. And although a computer-operated cattle prod might arouse serious interest, it might also provoke serious objections. Besides, how does the computer know what pizza parlor the human being went to???}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhjubeh}(h]h ]h"]h$]h&]uh1j hjsubeh}(h]h ]h"]h$]h&]uh1jq hhhMhjAhhubeh}(h]h ]h"]h$]h&]uh1jl hj?hhhhhNubh)}(h%:ref:`Back to Quick Quiz #3 `h]h)}(hjh]j)}(hjh]hBack to Quick Quiz #3}(hhhjhhhNhNubah}(h]h ](jstdstd-refeh"]h$]h&]uh1jhjubah}(h]h ]h"]h$]h&]refdocj refdomainjreftyperef refexplicitrefwarnjquiz_3uh1hhhhMhjubah}(h]h ]h"]h$]h&]uh1hhhhMhj?hhubh)}(hACKNOWLEDGEMENTSh]hACKNOWLEDGEMENTS}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj?hhubh)}(hMy thanks to the people who helped make this human-readable, including Jon Walpole, Josh Triplett, Serge Hallyn, Suzanne Wood, and Alan Stern.h]hMy thanks to the people who helped make this human-readable, including Jon Walpole, Josh Triplett, Serge Hallyn, Suzanne Wood, and Alan Stern.}(hjhjhhhNhNubah}(h]h ]h"]h$]h&]uh1hhhhMhj?hhubh)}(hAFor more information, see http://www.rdrop.com/users/paulmck/RCU.h](hFor more information, see }(hFor more information, see hjhhhNhNubh)}(h&http://www.rdrop.com/users/paulmck/RCUh]h&http://www.rdrop.com/users/paulmck/RCU}(hhhjhhhNhNubah}(h]h ]h"]h$]h&]refurijuh1hhjubh.}(hjhjhhhNhNubeh}(h]h ]h"]h$]h&]uh1hhhhMhj?hhubeh}(h](answers-to-quick-quizzesj3eh ]h"](9. answers to quick quizzes 9_whatisrcueh$]h&]uh1hhhhhhhhMj}j#j)sj}j3j)subeh}(h](what-is-rcu-read-copy-updateheh ]h"]($what is rcu? -- "read, copy, update" whatisrcu_doceh$]h&]uh1hhhhhhhhKj}j.hsj}hhsubeh}(h]h ]h"]h$]h&]sourcehuh1hcurrent_sourceN current_lineNsettingsdocutils.frontendValues)}(hN generatorN datestampN source_linkN source_urlN toc_backlinksentryfootnote_backlinksK sectnum_xformKstrip_commentsNstrip_elements_with_classesN strip_classesN report_levelK halt_levelKexit_status_levelKdebugNwarning_streamN tracebackinput_encoding utf-8-siginput_encoding_error_handlerstrictoutput_encodingutf-8output_encoding_error_handlerjVerror_encodingUTF-8error_encoding_error_handlerbackslashreplace language_codeenrecord_dependenciesNconfigN id_prefixhauto_id_prefixid dump_settingsNdump_internalsNdump_transformsNdump_pseudo_xmlNexpose_internalsNstrict_visitorN_disable_configN_sourceh _destinationN _config_files]7/var/lib/git/docbuild/linux/Documentation/docutils.confapep_referencesN pep_base_urlhttps://peps.python.org/pep_file_url_templatepep-%04drfc_referencesN rfc_base_url&https://datatracker.ietf.org/doc/html/ tab_widthKtrim_footnote_reference_spacefile_insertion_enabled raw_enabledKline_length_limitM'syntax_highlightlong smart_quotessmartquotes_locales]character_level_inline_markupdoctitle_xform docinfo_xformKsectsubtitle_xform image_loadinglinkembed_stylesheetcloak_email_addressessection_self_link embed_imagesenvNubreporterNindirect_targets]substitution_defs}substitution_names}refnames}(1]ja2]jlaurefids}(h]haj]jaj]jaj ]j aj ]j ajm ]jc ajk ]ja aj]jaj]jaj]jaj9]j/aj]jaj3]j)aunameids}(j.hj-j*jjjjj jj j j jjUjRjjjjj j jj&jj{j j j j jt j js jp jjm jjj j j jk jjjjjRjj@jj?j<jj9jjj:jj9j6j#j3j"ju nametypes}(j.j-NjjNj j Nj NjUNjNjNj Njjj j Njt js NjjNj Nj jNjjRj@j?NjjNj:j9Nj#j"Nuh}(hhj*hjjjjjjj jjjjRj jjXjjj jj!jjvjlj&jj{jj j j j j j jp j jm jy jjy j j jk jn jj jjjjjjj<jj9jEjjEjjj6jj3j?jj?u footnote_refs}(j]jaj]jlau citation_refs} autofootnotes]autofootnote_refs]symbol_footnotes]symbol_footnote_refs] footnotes](jje citations]autofootnote_startKsymbol_footnote_startK id_counter collectionsCounter}jdK sRparse_messages]transform_messages](hsystem_message)}(hhh]h)}(hhh]h3Hyperlink target "whatisrcu-doc" is not referenced.}(hhhjubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypeINFOsourcehlineKuh1jubj)}(hhh]h)}(hhh]h/Hyperlink target "whatisrcu" is not referenced.}(hhhjubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineKGuh1jubj)}(hhh]h)}(hhh]h)Hyperlink target "id1" is not referenced.}(hhhjubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineKuh1jubj)}(hhh]h)}(hhh]h)Hyperlink target "id6" is not referenced.}(hhhj"ubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jubj)}(hhh]h)}(hhh]h)Hyperlink target "id7" is not referenced.}(hhhj<ubah}(h]h ]h"]h$]h&]uh1hhj9ubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jubj)}(hhh]h)}(hhh]h)Hyperlink target "id8" is not referenced.}(hhhjVubah}(h]h ]h"]h$]h&]uh1hhjSubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMjuh1jubj)}(hhh]h)}(hhh]h,Hyperlink target "quiz-1" is not referenced.}(hhhjpubah}(h]h ]h"]h$]h&]uh1hhjmubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jubj)}(hhh]h)}(hhh]h,Hyperlink target "quiz-2" is not referenced.}(hhhjubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jubj)}(hhh]h)}(hhh]h,Hyperlink target "quiz-3" is not referenced.}(hhhjubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jubj)}(hhh]h)}(hhh]h)Hyperlink target "id9" is not referenced.}(hhhjubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jubj)}(hhh]h)}(hhh]h*Hyperlink target "id10" is not referenced.}(hhhjubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jubj)}(hhh]h)}(hhh]h*Hyperlink target "id11" is not referenced.}(hhhjubah}(h]h ]h"]h$]h&]uh1hhjubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jubj)}(hhh]h)}(hhh]h*Hyperlink target "id12" is not referenced.}(hhhj ubah}(h]h ]h"]h$]h&]uh1hhj ubah}(h]h ]h"]h$]h&]levelKtypejsourcehlineMuh1jube transformerN include_log] decorationNhhub.