diff options
Diffstat (limited to '')
61 files changed, 11281 insertions, 0 deletions
diff --git a/src/dmclock/.gitignore b/src/dmclock/.gitignore new file mode 100644 index 000000000..db9226e3c --- /dev/null +++ b/src/dmclock/.gitignore @@ -0,0 +1,5 @@ +*~ +*.dSYM +*.o +build* +cscope.* diff --git a/src/dmclock/.travis.yml b/src/dmclock/.travis.yml new file mode 100644 index 000000000..8fb8f898e --- /dev/null +++ b/src/dmclock/.travis.yml @@ -0,0 +1,29 @@ +language: cpp +cache: ccache +dist: xenial +sudo: false +branches: + only: + - master +os: + - linux +compiler: + - clang + - gcc +addons: + apt: + sources: + - ubuntu-toolchain-r-test + packages: + - cmake + - libgtest-dev + - libboost-dev +before_script: + - mkdir build + - cd build +script: + - cmake .. + - cmake --build . -- -j2 && ctest -V -j2 +env: + global: + - LANG="en_US.UTF-8" diff --git a/src/dmclock/CMakeLists.txt b/src/dmclock/CMakeLists.txt new file mode 100644 index 000000000..049ea5e31 --- /dev/null +++ b/src/dmclock/CMakeLists.txt @@ -0,0 +1,44 @@ +cmake_minimum_required(VERSION 3.5.1) + +project(dmclock CXX) + +list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_LIST_DIR}/cmake/modules") + +if (NOT(BOOST_FOUND)) + find_package(Boost REQUIRED) +endif() + +find_package(Threads) + +if(CMAKE_CXX_STANDARD OR CMAKE_CXX_FLAGS MATCHES "-std=(c|gnu)\\+\\+") + # use existing settings if available +else() + set(CMAKE_CXX_STANDARD 11) + set(CMAKE_CXX_STANDARD_REQUIRED ON) +endif() + +add_subdirectory(src) + +# Determine if dmclock is built as a subproject (using add_subdirectory) +# or if it is the master project. +set(MASTER_PROJECT FALSE) +if(CMAKE_CURRENT_SOURCE_DIR STREQUAL CMAKE_SOURCE_DIR) + set(MASTER_PROJECT TRUE) +endif() + +option(dmclock_TEST "Generate test targets" ${MASTER_PROJECT}) +if(dmclock_TEST) + if (NOT(TARGET gtest AND TARGET gtest_main)) + if (NOT GTEST_FOUND) + find_package(GTest QUIET) + if (NOT GTEST_FOUND) + include(BuildGTest) + endif() + endif() + endif() + + enable_testing() + add_subdirectory(test) + add_subdirectory(support/test) + add_subdirectory(sim) +endif() diff --git a/src/dmclock/COPYING b/src/dmclock/COPYING new file mode 100644 index 000000000..6f1dfff1f --- /dev/null +++ b/src/dmclock/COPYING @@ -0,0 +1,3 @@ +Files: * +Copyright: (C) 2016-2018 by Red Hat Inc. +License: LGPL2.1 (see COPYING-LGPL2.1) diff --git a/src/dmclock/COPYING-LGPL2.1 b/src/dmclock/COPYING-LGPL2.1 new file mode 100644 index 000000000..5ab7695ab --- /dev/null +++ b/src/dmclock/COPYING-LGPL2.1 @@ -0,0 +1,504 @@ + GNU LESSER GENERAL PUBLIC LICENSE + Version 2.1, February 1999 + + Copyright (C) 1991, 1999 Free Software Foundation, Inc. + 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + Everyone is permitted to copy and distribute verbatim copies + of this license document, but changing it is not allowed. + +[This is the first released version of the Lesser GPL. It also counts + as the successor of the GNU Library Public License, version 2, hence + the version number 2.1.] + + Preamble + + The licenses for most software are designed to take away your +freedom to share and change it. By contrast, the GNU General Public +Licenses are intended to guarantee your freedom to share and change +free software--to make sure the software is free for all its users. + + This license, the Lesser General Public License, applies to some +specially designated software packages--typically libraries--of the +Free Software Foundation and other authors who decide to use it. You +can use it too, but we suggest you first think carefully about whether +this license or the ordinary General Public License is the better +strategy to use in any particular case, based on the explanations below. + + When we speak of free software, we are referring to freedom of use, +not price. Our General Public Licenses are designed to make sure that +you have the freedom to distribute copies of free software (and charge +for this service if you wish); that you receive source code or can get +it if you want it; that you can change the software and use pieces of +it in new free programs; and that you are informed that you can do +these things. + + To protect your rights, we need to make restrictions that forbid +distributors to deny you these rights or to ask you to surrender these +rights. These restrictions translate to certain responsibilities for +you if you distribute copies of the library or if you modify it. + + For example, if you distribute copies of the library, whether gratis +or for a fee, you must give the recipients all the rights that we gave +you. You must make sure that they, too, receive or can get the source +code. If you link other code with the library, you must provide +complete object files to the recipients, so that they can relink them +with the library after making changes to the library and recompiling +it. And you must show them these terms so they know their rights. + + We protect your rights with a two-step method: (1) we copyright the +library, and (2) we offer you this license, which gives you legal +permission to copy, distribute and/or modify the library. + + To protect each distributor, we want to make it very clear that +there is no warranty for the free library. Also, if the library is +modified by someone else and passed on, the recipients should know +that what they have is not the original version, so that the original +author's reputation will not be affected by problems that might be +introduced by others. + + Finally, software patents pose a constant threat to the existence of +any free program. We wish to make sure that a company cannot +effectively restrict the users of a free program by obtaining a +restrictive license from a patent holder. Therefore, we insist that +any patent license obtained for a version of the library must be +consistent with the full freedom of use specified in this license. + + Most GNU software, including some libraries, is covered by the +ordinary GNU General Public License. This license, the GNU Lesser +General Public License, applies to certain designated libraries, and +is quite different from the ordinary General Public License. We use +this license for certain libraries in order to permit linking those +libraries into non-free programs. + + When a program is linked with a library, whether statically or using +a shared library, the combination of the two is legally speaking a +combined work, a derivative of the original library. The ordinary +General Public License therefore permits such linking only if the +entire combination fits its criteria of freedom. The Lesser General +Public License permits more lax criteria for linking other code with +the library. + + We call this license the "Lesser" General Public License because it +does Less to protect the user's freedom than the ordinary General +Public License. It also provides other free software developers Less +of an advantage over competing non-free programs. These disadvantages +are the reason we use the ordinary General Public License for many +libraries. However, the Lesser license provides advantages in certain +special circumstances. + + For example, on rare occasions, there may be a special need to +encourage the widest possible use of a certain library, so that it becomes +a de-facto standard. To achieve this, non-free programs must be +allowed to use the library. A more frequent case is that a free +library does the same job as widely used non-free libraries. In this +case, there is little to gain by limiting the free library to free +software only, so we use the Lesser General Public License. + + In other cases, permission to use a particular library in non-free +programs enables a greater number of people to use a large body of +free software. For example, permission to use the GNU C Library in +non-free programs enables many more people to use the whole GNU +operating system, as well as its variant, the GNU/Linux operating +system. + + Although the Lesser General Public License is Less protective of the +users' freedom, it does ensure that the user of a program that is +linked with the Library has the freedom and the wherewithal to run +that program using a modified version of the Library. + + The precise terms and conditions for copying, distribution and +modification follow. Pay close attention to the difference between a +"work based on the library" and a "work that uses the library". The +former contains code derived from the library, whereas the latter must +be combined with the library in order to run. + + GNU LESSER GENERAL PUBLIC LICENSE + TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION + + 0. This License Agreement applies to any software library or other +program which contains a notice placed by the copyright holder or +other authorized party saying it may be distributed under the terms of +this Lesser General Public License (also called "this License"). +Each licensee is addressed as "you". + + A "library" means a collection of software functions and/or data +prepared so as to be conveniently linked with application programs +(which use some of those functions and data) to form executables. + + The "Library", below, refers to any such software library or work +which has been distributed under these terms. A "work based on the +Library" means either the Library or any derivative work under +copyright law: that is to say, a work containing the Library or a +portion of it, either verbatim or with modifications and/or translated +straightforwardly into another language. (Hereinafter, translation is +included without limitation in the term "modification".) + + "Source code" for a work means the preferred form of the work for +making modifications to it. For a library, complete source code means +all the source code for all modules it contains, plus any associated +interface definition files, plus the scripts used to control compilation +and installation of the library. + + Activities other than copying, distribution and modification are not +covered by this License; they are outside its scope. The act of +running a program using the Library is not restricted, and output from +such a program is covered only if its contents constitute a work based +on the Library (independent of the use of the Library in a tool for +writing it). Whether that is true depends on what the Library does +and what the program that uses the Library does. + + 1. You may copy and distribute verbatim copies of the Library's +complete source code as you receive it, in any medium, provided that +you conspicuously and appropriately publish on each copy an +appropriate copyright notice and disclaimer of warranty; keep intact +all the notices that refer to this License and to the absence of any +warranty; and distribute a copy of this License along with the +Library. + + You may charge a fee for the physical act of transferring a copy, +and you may at your option offer warranty protection in exchange for a +fee. + + 2. You may modify your copy or copies of the Library or any portion +of it, thus forming a work based on the Library, and copy and +distribute such modifications or work under the terms of Section 1 +above, provided that you also meet all of these conditions: + + a) The modified work must itself be a software library. + + b) You must cause the files modified to carry prominent notices + stating that you changed the files and the date of any change. + + c) You must cause the whole of the work to be licensed at no + charge to all third parties under the terms of this License. + + d) If a facility in the modified Library refers to a function or a + table of data to be supplied by an application program that uses + the facility, other than as an argument passed when the facility + is invoked, then you must make a good faith effort to ensure that, + in the event an application does not supply such function or + table, the facility still operates, and performs whatever part of + its purpose remains meaningful. + + (For example, a function in a library to compute square roots has + a purpose that is entirely well-defined independent of the + application. Therefore, Subsection 2d requires that any + application-supplied function or table used by this function must + be optional: if the application does not supply it, the square + root function must still compute square roots.) + +These requirements apply to the modified work as a whole. If +identifiable sections of that work are not derived from the Library, +and can be reasonably considered independent and separate works in +themselves, then this License, and its terms, do not apply to those +sections when you distribute them as separate works. But when you +distribute the same sections as part of a whole which is a work based +on the Library, the distribution of the whole must be on the terms of +this License, whose permissions for other licensees extend to the +entire whole, and thus to each and every part regardless of who wrote +it. + +Thus, it is not the intent of this section to claim rights or contest +your rights to work written entirely by you; rather, the intent is to +exercise the right to control the distribution of derivative or +collective works based on the Library. + +In addition, mere aggregation of another work not based on the Library +with the Library (or with a work based on the Library) on a volume of +a storage or distribution medium does not bring the other work under +the scope of this License. + + 3. You may opt to apply the terms of the ordinary GNU General Public +License instead of this License to a given copy of the Library. To do +this, you must alter all the notices that refer to this License, so +that they refer to the ordinary GNU General Public License, version 2, +instead of to this License. (If a newer version than version 2 of the +ordinary GNU General Public License has appeared, then you can specify +that version instead if you wish.) Do not make any other change in +these notices. + + Once this change is made in a given copy, it is irreversible for +that copy, so the ordinary GNU General Public License applies to all +subsequent copies and derivative works made from that copy. + + This option is useful when you wish to copy part of the code of +the Library into a program that is not a library. + + 4. You may copy and distribute the Library (or a portion or +derivative of it, under Section 2) in object code or executable form +under the terms of Sections 1 and 2 above provided that you accompany +it with the complete corresponding machine-readable source code, which +must be distributed under the terms of Sections 1 and 2 above on a +medium customarily used for software interchange. + + If distribution of object code is made by offering access to copy +from a designated place, then offering equivalent access to copy the +source code from the same place satisfies the requirement to +distribute the source code, even though third parties are not +compelled to copy the source along with the object code. + + 5. A program that contains no derivative of any portion of the +Library, but is designed to work with the Library by being compiled or +linked with it, is called a "work that uses the Library". Such a +work, in isolation, is not a derivative work of the Library, and +therefore falls outside the scope of this License. + + However, linking a "work that uses the Library" with the Library +creates an executable that is a derivative of the Library (because it +contains portions of the Library), rather than a "work that uses the +library". The executable is therefore covered by this License. +Section 6 states terms for distribution of such executables. + + When a "work that uses the Library" uses material from a header file +that is part of the Library, the object code for the work may be a +derivative work of the Library even though the source code is not. +Whether this is true is especially significant if the work can be +linked without the Library, or if the work is itself a library. The +threshold for this to be true is not precisely defined by law. + + If such an object file uses only numerical parameters, data +structure layouts and accessors, and small macros and small inline +functions (ten lines or less in length), then the use of the object +file is unrestricted, regardless of whether it is legally a derivative +work. (Executables containing this object code plus portions of the +Library will still fall under Section 6.) + + Otherwise, if the work is a derivative of the Library, you may +distribute the object code for the work under the terms of Section 6. +Any executables containing that work also fall under Section 6, +whether or not they are linked directly with the Library itself. + + 6. As an exception to the Sections above, you may also combine or +link a "work that uses the Library" with the Library to produce a +work containing portions of the Library, and distribute that work +under terms of your choice, provided that the terms permit +modification of the work for the customer's own use and reverse +engineering for debugging such modifications. + + You must give prominent notice with each copy of the work that the +Library is used in it and that the Library and its use are covered by +this License. You must supply a copy of this License. If the work +during execution displays copyright notices, you must include the +copyright notice for the Library among them, as well as a reference +directing the user to the copy of this License. Also, you must do one +of these things: + + a) Accompany the work with the complete corresponding + machine-readable source code for the Library including whatever + changes were used in the work (which must be distributed under + Sections 1 and 2 above); and, if the work is an executable linked + with the Library, with the complete machine-readable "work that + uses the Library", as object code and/or source code, so that the + user can modify the Library and then relink to produce a modified + executable containing the modified Library. (It is understood + that the user who changes the contents of definitions files in the + Library will not necessarily be able to recompile the application + to use the modified definitions.) + + b) Use a suitable shared library mechanism for linking with the + Library. A suitable mechanism is one that (1) uses at run time a + copy of the library already present on the user's computer system, + rather than copying library functions into the executable, and (2) + will operate properly with a modified version of the library, if + the user installs one, as long as the modified version is + interface-compatible with the version that the work was made with. + + c) Accompany the work with a written offer, valid for at + least three years, to give the same user the materials + specified in Subsection 6a, above, for a charge no more + than the cost of performing this distribution. + + d) If distribution of the work is made by offering access to copy + from a designated place, offer equivalent access to copy the above + specified materials from the same place. + + e) Verify that the user has already received a copy of these + materials or that you have already sent this user a copy. + + For an executable, the required form of the "work that uses the +Library" must include any data and utility programs needed for +reproducing the executable from it. However, as a special exception, +the materials to be distributed need not include anything that is +normally distributed (in either source or binary form) with the major +components (compiler, kernel, and so on) of the operating system on +which the executable runs, unless that component itself accompanies +the executable. + + It may happen that this requirement contradicts the license +restrictions of other proprietary libraries that do not normally +accompany the operating system. Such a contradiction means you cannot +use both them and the Library together in an executable that you +distribute. + + 7. You may place library facilities that are a work based on the +Library side-by-side in a single library together with other library +facilities not covered by this License, and distribute such a combined +library, provided that the separate distribution of the work based on +the Library and of the other library facilities is otherwise +permitted, and provided that you do these two things: + + a) Accompany the combined library with a copy of the same work + based on the Library, uncombined with any other library + facilities. This must be distributed under the terms of the + Sections above. + + b) Give prominent notice with the combined library of the fact + that part of it is a work based on the Library, and explaining + where to find the accompanying uncombined form of the same work. + + 8. You may not copy, modify, sublicense, link with, or distribute +the Library except as expressly provided under this License. Any +attempt otherwise to copy, modify, sublicense, link with, or +distribute the Library is void, and will automatically terminate your +rights under this License. However, parties who have received copies, +or rights, from you under this License will not have their licenses +terminated so long as such parties remain in full compliance. + + 9. You are not required to accept this License, since you have not +signed it. However, nothing else grants you permission to modify or +distribute the Library or its derivative works. These actions are +prohibited by law if you do not accept this License. Therefore, by +modifying or distributing the Library (or any work based on the +Library), you indicate your acceptance of this License to do so, and +all its terms and conditions for copying, distributing or modifying +the Library or works based on it. + + 10. Each time you redistribute the Library (or any work based on the +Library), the recipient automatically receives a license from the +original licensor to copy, distribute, link with or modify the Library +subject to these terms and conditions. You may not impose any further +restrictions on the recipients' exercise of the rights granted herein. +You are not responsible for enforcing compliance by third parties with +this License. + + 11. If, as a consequence of a court judgment or allegation of patent +infringement or for any other reason (not limited to patent issues), +conditions are imposed on you (whether by court order, agreement or +otherwise) that contradict the conditions of this License, they do not +excuse you from the conditions of this License. If you cannot +distribute so as to satisfy simultaneously your obligations under this +License and any other pertinent obligations, then as a consequence you +may not distribute the Library at all. For example, if a patent +license would not permit royalty-free redistribution of the Library by +all those who receive copies directly or indirectly through you, then +the only way you could satisfy both it and this License would be to +refrain entirely from distribution of the Library. + +If any portion of this section is held invalid or unenforceable under any +particular circumstance, the balance of the section is intended to apply, +and the section as a whole is intended to apply in other circumstances. + +It is not the purpose of this section to induce you to infringe any +patents or other property right claims or to contest validity of any +such claims; this section has the sole purpose of protecting the +integrity of the free software distribution system which is +implemented by public license practices. Many people have made +generous contributions to the wide range of software distributed +through that system in reliance on consistent application of that +system; it is up to the author/donor to decide if he or she is willing +to distribute software through any other system and a licensee cannot +impose that choice. + +This section is intended to make thoroughly clear what is believed to +be a consequence of the rest of this License. + + 12. If the distribution and/or use of the Library is restricted in +certain countries either by patents or by copyrighted interfaces, the +original copyright holder who places the Library under this License may add +an explicit geographical distribution limitation excluding those countries, +so that distribution is permitted only in or among countries not thus +excluded. In such case, this License incorporates the limitation as if +written in the body of this License. + + 13. The Free Software Foundation may publish revised and/or new +versions of the Lesser General Public License from time to time. +Such new versions will be similar in spirit to the present version, +but may differ in detail to address new problems or concerns. + +Each version is given a distinguishing version number. If the Library +specifies a version number of this License which applies to it and +"any later version", you have the option of following the terms and +conditions either of that version or of any later version published by +the Free Software Foundation. If the Library does not specify a +license version number, you may choose any version ever published by +the Free Software Foundation. + + 14. If you wish to incorporate parts of the Library into other free +programs whose distribution conditions are incompatible with these, +write to the author to ask for permission. For software which is +copyrighted by the Free Software Foundation, write to the Free +Software Foundation; we sometimes make exceptions for this. Our +decision will be guided by the two goals of preserving the free status +of all derivatives of our free software and of promoting the sharing +and reuse of software generally. + + NO WARRANTY + + 15. BECAUSE THE LIBRARY IS LICENSED FREE OF CHARGE, THERE IS NO +WARRANTY FOR THE LIBRARY, TO THE EXTENT PERMITTED BY APPLICABLE LAW. +EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR +OTHER PARTIES PROVIDE THE LIBRARY "AS IS" WITHOUT WARRANTY OF ANY +KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR +PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE +LIBRARY IS WITH YOU. SHOULD THE LIBRARY PROVE DEFECTIVE, YOU ASSUME +THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. + + 16. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN +WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY +AND/OR REDISTRIBUTE THE LIBRARY AS PERMITTED ABOVE, BE LIABLE TO YOU +FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR +CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE +LIBRARY (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING +RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A +FAILURE OF THE LIBRARY TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF +SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH +DAMAGES. + + END OF TERMS AND CONDITIONS + + How to Apply These Terms to Your New Libraries + + If you develop a new library, and you want it to be of the greatest +possible use to the public, we recommend making it free software that +everyone can redistribute and change. You can do so by permitting +redistribution under these terms (or, alternatively, under the terms of the +ordinary General Public License). + + To apply these terms, attach the following notices to the library. It is +safest to attach them to the start of each source file to most effectively +convey the exclusion of warranty; and each file should have at least the +"copyright" line and a pointer to where the full notice is found. + + <one line to give the library's name and a brief idea of what it does.> + Copyright (C) <year> <name of author> + + This library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + This library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with this library; if not, write to the Free Software + Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + +Also add information on how to contact you by electronic and paper mail. + +You should also get your employer (if you work as a programmer) or your +school, if any, to sign a "copyright disclaimer" for the library, if +necessary. Here is a sample; alter the names: + + Yoyodyne, Inc., hereby disclaims all copyright interest in the + library `Frob' (a library for tweaking knobs) written by James Random Hacker. + + <signature of Ty Coon>, 1 April 1990 + Ty Coon, President of Vice + +That's all there is to it! + + diff --git a/src/dmclock/README.md b/src/dmclock/README.md new file mode 100644 index 000000000..b046fd3e3 --- /dev/null +++ b/src/dmclock/README.md @@ -0,0 +1,45 @@ +# dmclock + +This repository contains C++ 11 code that implements the dmclock +distributed quality of service algorithm. See __mClock: Handling +Throughput Variability for Hypervisor IO Scheduling__ by Gulati, +Merchant, and Varman for a description of the algorithm. + +## Bugs and features + +There is a dmclock project under https://tracker.ceph.com/ through +which bugs can be reported and featuers requested. + +## Running cmake + +When running cmake, set the build type with either: + + -DCMAKE_BUILD_TYPE=Debug + -DCMAKE_BUILD_TYPE=Release + +To turn on profiling, run cmake with an additional: + + -DPROFILE=yes + +## Running make + +### Building the dmclock library + +The `make` command builds a library libdmclock.a. That plus the header +files in the src directory allow one to use the implementation in +their code. + +### Building unit tests + +The `make dmclock-tests` command builds unit tests. + +### Building simulations + +The `make dmclock-sims` command builds two simulations -- *dmc_sim* +and *ssched_sim* -- which incorporate, respectively, the dmclock +priority queue or a very simple scheduler for comparison. Other +priority queue implementations could be added in the future. + +## dmclock API + +To be written.... diff --git a/src/dmclock/benchmark/README.md b/src/dmclock/benchmark/README.md new file mode 100644 index 000000000..d945e986f --- /dev/null +++ b/src/dmclock/benchmark/README.md @@ -0,0 +1,42 @@ +# dmclock benchmarking + +**IMPORTANT**: now that K_WAY_HEAP is no longer allowed to have the +value 1, the shell and Python scripts that generate the PDFs no longer +work exactly correctly. Some effort to debug is necessary. + +This directory contains scripts to evaluate effects of different +branching-factors (k=1 to k=11) in the IndirectIntrusiveHeap +data-structure. IndirectIntrusiveHeap is now a k-way heap, so finding +an ideal value for k (i.e., k=2 or k=3) for a particular work-load is +important. Also, it is well-documented that the right choice of +k-value improves the caching behaviour [Syed -- citation needed +here]. As a result, the overall performance of an application using +k-way heap increases significantly [Syed -- citation needed here]. + +A rule of thumb is the following: + if number of elements are <= 6, use k=1 + otherwise, use k=3. + +## Prerequisites + +requires python 2.7, gnuplot, and awk. + +## Running benchmark + +./run.sh [name_of_the_output] [k_way] [repeat] # [Syed -- last two command line args do not work] + +The "run.sh" script looks for config files in the "configs" directory, +and the final output is generated as +"name_of_the_output.pdf". Internally, "run.sh" calls other scripts +such as data_gen.sh, data_parser.py, and plot_gen.sh. + +## Modifying parameters + +To modify k-value and/or the amount of times each simulation is +repeated, modify the following two variables in "run.sh" file: + + k_way=[your_value] + repeat=[your_value] + +For example, k_way=3 means, the benchmark will compare simulations +using 1-way, 2-way, and 3-way heaps. diff --git a/src/dmclock/benchmark/configs/dmc_sim_100_100.conf b/src/dmclock/benchmark/configs/dmc_sim_100_100.conf new file mode 100644 index 000000000..c93d4c71f --- /dev/null +++ b/src/dmclock/benchmark/configs/dmc_sim_100_100.conf @@ -0,0 +1,31 @@ +[global] +server_groups = 1 +client_groups = 2 +server_random_selection = true +server_soft_limit = true + +[server.0] +server_count = 100 +server_iops = 160 + +[client.0] +client_count = 99 +client_wait = 0 +client_total_ops = 10000 +client_server_select_range = 100 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 100.0 +client_limit = 0.0 +client_weight = 1.0 + +[client.1] +client_count = 1 +client_wait = 10 +client_total_ops = 10000 +client_server_select_range = 100 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 100.0 +client_limit = 0.0 +client_weight = 1.0 diff --git a/src/dmclock/benchmark/configs/dmc_sim_8_6.conf b/src/dmclock/benchmark/configs/dmc_sim_8_6.conf new file mode 100644 index 000000000..28aeb401d --- /dev/null +++ b/src/dmclock/benchmark/configs/dmc_sim_8_6.conf @@ -0,0 +1,43 @@ +[global] +server_groups = 1 +client_groups = 3 +server_random_selection = true +server_soft_limit = true + +[client.0] +client_count = 2 +client_wait = 0 +client_total_ops = 1000 +client_server_select_range = 8 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 0.0 +client_limit = 0.0 +client_weight = 1.0 + +[client.1] +client_count = 2 +client_wait = 5 +client_total_ops = 1000 +client_server_select_range = 8 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 20.0 +client_limit = 40.0 +client_weight = 1.0 + +[client.2] +client_count = 2 +client_wait = 10 +client_total_ops = 1000 +client_server_select_range = 8 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 0.0 +client_limit = 50.0 +client_weight = 2.0 + + +[server.0] +server_count = 8 +server_iops = 160 diff --git a/src/dmclock/benchmark/data_gen.sh b/src/dmclock/benchmark/data_gen.sh new file mode 100755 index 000000000..80a77bd9a --- /dev/null +++ b/src/dmclock/benchmark/data_gen.sh @@ -0,0 +1,73 @@ +#!/bin/bash +config_dir="configs" +repeat=2 #5 + +# parameter check -- output_file name +if [ "$1" != "" ]; then + output_file="$1" +else + echo "Please provide the name of the output file" + exit +fi + +# parameter check -- k-value +if [ "$2" != "" ]; then + k_way="$2" +else + echo "Please provide the maximum K_WAY value" + exit +fi + +# parameter check --repeat +if [ "$3" != "" ]; then + repeat="$3" +fi + +echo "k-way:$k_way, num_repeat:$repeat" + +# create simulators in different directories +k=2 +while [ $k -le $k_way ] +do + mkdir "build_$k" + cd "build_$k" + rm -rf * + cmake -DCMAKE_BUILD_TYPE=Release -DK_WAY_HEAP=$k ../../. + make dmclock-sims + cd .. + + k=$(( $k + 1 )) +done + +# run simulators +echo '' > $output_file +for config in "$config_dir"/*.conf +do + k=2 + while [ $k -le $k_way ] + do + cd "build_$k" + + # repeat same experiment + i=0 + while [ $i -lt $repeat ] + do + i=$(( $i + 1 )) + + # clear cache first + sync + #sudo sh -c 'echo 1 >/proc/sys/vm/drop_caches' + #sudo sh -c 'echo 2 >/proc/sys/vm/drop_caches' + #sudo sh -c 'echo 3 >/proc/sys/vm/drop_caches' + + # run with heap + msg="file_name:$k:$config" + echo $msg >> ../$output_file + echo "running $msg ..." + ./sim/dmc_sim -c ../$config | awk '(/average/)' >> ../$output_file + done # end repeat + cd .. + k=$(( $k + 1 )) + done # end k_way +done # end config + diff --git a/src/dmclock/benchmark/data_parser.py b/src/dmclock/benchmark/data_parser.py new file mode 100755 index 000000000..c90d85fd9 --- /dev/null +++ b/src/dmclock/benchmark/data_parser.py @@ -0,0 +1,191 @@ +#!/usr/bin/env python + +class DataPoint: + def __init__(self): + self.nserver = 0; + self.nclient = 0; + self.heap_type = 0; + self.total_time_to_add_req = 0; + self.total_time_to_complete_req = 0; + self.config = '' + + def set_name(self, config, heap_type): + self.config = config; + self.heap_type = heap_type + + def get_conig(self): + import re + return re.split(r"/|\.", self.config)[1] + + def __str__(self): + return "s:%d, c:%d,h:%d,config:%s"%(self.nserver, self.nclient, self.heap_type, self.config); +# end DataPoint + + +def isFloat(elem): + try: + float(elem) + return True + except ValueError: + return False +#end isFloat + + +def parse_config_params(fname): + nclient = 0; + nserver = 0; + # read config file property + with open(fname, 'r') as f: + for line in f: + line = line.strip('\n \t') + if not line: continue; + if line.startswith("client_count"): + nclient += int(line.split('=')[-1]); + if line.startswith("server_count"): + nserver += int(line.split('=')[-1]); + # end of file + return [nserver, nclient]; +# parse_config_params + +def make_aggregate_data_point(dps, config, heap_type): + # create new aggregate point + dp = DataPoint(); + # set set and k_way_heap property + dp.set_name(config, heap_type); + + num_run = 0 + for _dp in dps: + if _dp.config == config and _dp.heap_type == heap_type: + # print _dp, config, heap_type + dp.nserver =_dp.nserver + dp.nclient = _dp.nclient + num_run += 1 + dp.total_time_to_add_req += _dp.total_time_to_add_req + dp.total_time_to_complete_req += _dp.total_time_to_complete_req + + # average + dp.total_time_to_add_req /= num_run; + dp.total_time_to_complete_req /= num_run + #print dp + return dp; + +def parse_data_points(filename): + dps = []; #data-points + dp = None; + state = 0; + configs = {} + k_ways = {} + + with open(filename, 'r') as f: + for line in f: + line = line.strip('\n \t') + if not line: continue; + + # file_name:1:configs/dmc_sim_8_6.conf + if line.startswith("file_name"): + if dp: + dps.append(dp); + state = 0; + + # new data-point + dp = DataPoint(); + parts = line.split(':') + fname = parts[-1]; + dp.heap_type = int(parts[1]); + if dp.heap_type not in k_ways: + k_ways[dp.heap_type] = 1; + + # add to the dictionary + configs[fname] = 1; + + dp.config = fname; + params = parse_config_params(fname) + dp.nserver = params[0]; + dp.nclient = params[-1]; + + elif line.startswith("average"): # take last 2 averages + r = [float(s) for s in line.split(' ') if isFloat(s)] + state +=1; + #print r, dp #if isFloat(s) + if state == 3: + dp.total_time_to_add_req = r[0] + elif state == 4: + dp.total_time_to_complete_req = r[0] + else: pass + + else: + pass; + # final entry + dps.append(dp) + + # compute average of multiple runs + dps_avg = [] + for config in configs: + data_per_config = [] + for k in k_ways: + aggr_dp = make_aggregate_data_point(dps, config , k); + data_per_config.append(aggr_dp); + dps_avg.append(data_per_config); + # end for + return dps_avg; +# end parse_data_points + + +def create_header(num_cols): + fields = ['nserver_nclient(config_file)','add_req', 'complete_req']; + header = fields[0] + #write add_req_{1, ...} + for i in range(num_cols): + header = '%s %s_%i'%(header, fields[1], i+2) + #write complete_req_{1, ...} + for i in range(num_cols): + header = '%s %s_%i'%(header, fields[2], i+2) + # new-line + header = '%s\n'%(header) + return header +# end create_header + + +def create_data_line(aggr_dp): + # get common info + dp = aggr_dp[0] + data_line = "s:%d_c:%d "%(dp.nserver, dp.nclient); + # get the point-count + num_cols = len(aggr_dp); + # write add_req_{1, ...} + for i in range(num_cols): + data_line = '%s %f'%(data_line, aggr_dp[i].total_time_to_add_req) + # write complete_req_{1, ...} + for i in range(num_cols): + data_line = '%s %f'%(data_line, aggr_dp[i].total_time_to_complete_req) + # new-line + data_line = '%s\n'%(data_line) + return data_line +# end create_data_line + + +def make_data(filename): + # write the aggregated point in space separated file + dps = parse_data_points(filename); + if not len(dps) : return + print "total points: ", len(dps) + # open file + with open('%s.dat'%(filename), 'w+') as f: + # write header + f.write(create_header(len(dps[0]))); + # write data-line + for aggr_dp in dps: + f.write(create_data_line(aggr_dp)); + + +def main(output_file): + print output_file + make_data(output_file); + +import sys +if __name__ == "__main__": + file_name="result" + if len(sys.argv) > 1: + file_name=sys.argv[1].strip() + main(file_name) + diff --git a/src/dmclock/benchmark/plot_gen.sh b/src/dmclock/benchmark/plot_gen.sh new file mode 100755 index 000000000..d90bde192 --- /dev/null +++ b/src/dmclock/benchmark/plot_gen.sh @@ -0,0 +1,60 @@ +#!/bin/bash + +if [ "$1" != "" ]; then + output_file="$1" +else + echo "Please provide the name of the output file" + exit +fi + +# parameter check -- k-value +if [ "$2" != "" ]; then + k_way="$2" +else + echo "Please provide the maximum K_WAY value" + exit +fi +#echo "k-way: $k_way" +#exit + +gnuplot << EOF + +# Note you need gnuplot 4.4 for the pdfcairo terminal. +clear +reset + +set terminal pdfcairo size 7in,5in font "Gill Sans,5" linewidth 1 rounded fontscale .8 noenhanced +set output "${output_file}.pdf" + +# starts multiplot +set multiplot layout 2,1 + +# Line style for axes +set style line 80 lt rgb "#808080" + +# Line style for grid +set style line 81 lt 0 # dashed +set style line 81 lt rgb "#808080" # grey + +set grid back linestyle 81 +set border 3 back linestyle 80 + +#set xtics rotate out +set style data histogram +set style histogram clustered + +set style fill solid border +set xlabel 'Heap Timing for different K values' +set ylabel 'Time (nanosec)' +set key top right + +set yrange [0:*] + +# plot 1 +set title 'Request Addition Time' +plot for [COL=2:($k_way + 1)] '${output_file}.dat' using COL:xticlabels(1) title columnheader + +# plot 2 +set title 'Request Completion Time' +plot for [COL=($k_way + 2):(2 * $k_way + 1)] '${output_file}.dat' using COL:xticlabels(1) title columnheader +EOF diff --git a/src/dmclock/benchmark/run.sh b/src/dmclock/benchmark/run.sh new file mode 100755 index 000000000..11432b530 --- /dev/null +++ b/src/dmclock/benchmark/run.sh @@ -0,0 +1,24 @@ +#!/bin/bash + +# default value +k_way=3 #11 +repeat=2 #5 + +output_file="" +if [ "$1" != "" ]; then + output_file="$1" +else + echo "Please provide the name of the output file" + exit +fi + +echo "generating file ${output_file}" +sh data_gen.sh ${output_file} ${k_way} ${repeat} + +echo "converting ${output_file} to ${output_file}.dat" +python data_parser.py ${output_file} + +echo "now generating bar-chart" +#gnuplot -e 'output_file=value' plot_gen.gnuplot +sh plot_gen.sh ${output_file} ${k_way} +echo "done! check ${output_file}.pdf" diff --git a/src/dmclock/cmake/modules/BuildGTest.cmake b/src/dmclock/cmake/modules/BuildGTest.cmake new file mode 100644 index 000000000..2b3f2ae52 --- /dev/null +++ b/src/dmclock/cmake/modules/BuildGTest.cmake @@ -0,0 +1,71 @@ +macro(_build_gtest gtest_root) + include(ExternalProject) + ExternalProject_Add(googletest + SOURCE_DIR ${gtest_root} + CMAKE_ARGS -DBUILD_GMOCK=OFF -DBUILD_GTEST=ON + INSTALL_COMMAND "" + LOG_CONFIGURE ON + LOG_BUILD ON) + + ExternalProject_Get_Property(googletest source_dir) + find_path(GTEST_INCLUDE_DIRS + NAMES gtest/gtest.h + PATHS ${source_dir}/googletest/include /usr/include) + find_path(GMOCK_INCLUDE_DIRS + NAMES gmock/gmock.h + PATHS ${source_dir}/googlemock/include /usr/include) + + find_package(Threads REQUIRED) + + ExternalProject_Get_Property(googletest binary_dir) + set(GTEST_LIBRARY_PATH ${binary_dir}/${CMAKE_FIND_LIBRARY_PREFIXES}gtest.a) + set(GTEST_LIBRARY GTest::GTest) + add_library(${GTEST_LIBRARY} STATIC IMPORTED) + set_target_properties(${GTEST_LIBRARY} PROPERTIES + INTERFACE_INCLUDE_DIRECTORIES "${GTEST_INCLUDE_DIRS}" + IMPORTED_LOCATION ${GTEST_LIBRARY_PATH} + IMPORTED_LINK_INTERFACE_LANGUAGES "CXX" + IMPORTED_LINK_INTERFACE_LIBRARIES ${CMAKE_THREAD_LIBS_INIT}) + add_dependencies(${GTEST_LIBRARY} googletest) + set(GTEST_LIBRARIES ${GTEST_LIBRARY}) + + set(GTEST_MAIN_LIBRARY_PATH ${binary_dir}/${CMAKE_FIND_LIBRARY_PREFIXES}gtest_main.a) + set(GTEST_MAIN_LIBRARY GTest::Main) + add_library(${GTEST_MAIN_LIBRARY} STATIC IMPORTED) + set_target_properties(${GTEST_MAIN_LIBRARY} PROPERTIES + INTERFACE_INCLUDE_DIRECTORIES "${GTEST_INCLUDE_DIRS}" + IMPORTED_LOCATION ${GTEST_MAIN_LIBRARY_PATH} + IMPORTED_LINK_INTERFACE_LIBRARIES ${CMAKE_THREAD_LIBS_INIT}) + add_dependencies(${GTEST_MAIN_LIBRARY} googletest) + + set(GMOCK_LIBRARY_PATH ${binary_dir}/${CMAKE_FIND_LIBRARY_PREFIXES}gmock.a) + set(GMOCK_LIBRARY GMock::GMock) + add_library(${GMOCK_LIBRARY} STATIC IMPORTED) + set_target_properties(${GMOCK_LIBRARY} PROPERTIES + INTERFACE_INCLUDE_DIRECTORIES "${GMOCK_INCLUDE_DIRS}" + IMPORTED_LOCATION "${GMOCK_LIBRARY_PATH}" + IMPORTED_LINK_INTERFACE_LANGUAGES "CXX" + IMPORTED_LINK_INTERFACE_LIBRARIES ${CMAKE_THREAD_LIBS_INIT}) + add_dependencies(${GMOCK_LIBRARY} googletest) + + set(GMOCK_MAIN_LIBRARY_PATH ${binary_dir}/${CMAKE_FIND_LIBRARY_PREFIXES}gmock_main.a) + set(GMOCK_MAIN_LIBRARY GMock::Main) + add_library(${GMOCK_MAIN_LIBRARY} STATIC IMPORTED) + set_target_properties(${GMOCK_MAIN_LIBRARY} PROPERTIES + INTERFACE_INCLUDE_DIRECTORIES "${GMOCK_INCLUDE_DIRS}" + IMPORTED_LOCATION ${GMOCK_MAIN_LIBRARY_PATH} + IMPORTED_LINK_INTERFACE_LANGUAGES "CXX" + IMPORTED_LINK_INTERFACE_LIBRARIES ${CMAKE_THREAD_LIBS_INIT}) + add_dependencies(${GMOCK_MAIN_LIBRARY} ${GTEST_LIBRARY}) +endmacro() + +find_path(GTEST_ROOT src/gtest.cc + HINTS $ENV{GTEST_ROOT} + PATHS /usr/src/googletest/googletest /usr/src/gtest) + +if(EXISTS ${GTEST_ROOT}) + message(STATUS "Found googletest: ${GTEST_ROOT}") + _build_gtest(${GTEST_ROOT}) +else() + message(SEND_ERROR "Could NOT find googletest") +endif() diff --git a/src/dmclock/dmclock-config.cmake.in b/src/dmclock/dmclock-config.cmake.in new file mode 100644 index 000000000..01636532c --- /dev/null +++ b/src/dmclock/dmclock-config.cmake.in @@ -0,0 +1,17 @@ +# - Config file for the FooBar package +# It defines the following variables +# DMCLOCK_INCLUDE_DIRS - include directories for FooBar +# DMCLOCK_LIBRARIES - libraries to link against + +# Compute paths +get_filename_component(DMCLOCK_CMAKE_DIR "${CMAKE_CURRENT_LIST_FILE}" PATH) +set(DMCLOCK_INCLUDE_DIRS "${DMCLOCK_CMAKE_DIR}/src") +# set(DMCLOCK_INCLUDE_DIRS "@CONF_INCLUDE_DIRS@") + +# Our library dependencies (contains definitions for IMPORTED targets) +if(NOT TARGET dmclock AND NOT dmclock_BINARY_DIR) + include("${DMCLOCK_CMAKE_DIR}/dmclock-targets.cmake") +endif() + +# These are IMPORTED targets created by FooBarTargets.cmake +set(DMCLOCK_LIBRARIES dmclock) diff --git a/src/dmclock/dmclock-targets.cmake b/src/dmclock/dmclock-targets.cmake new file mode 100644 index 000000000..2c84f34a1 --- /dev/null +++ b/src/dmclock/dmclock-targets.cmake @@ -0,0 +1 @@ +export(PACKAGE dmclock) diff --git a/src/dmclock/sim/CMakeLists.txt b/src/dmclock/sim/CMakeLists.txt new file mode 100644 index 000000000..ca537ce6f --- /dev/null +++ b/src/dmclock/sim/CMakeLists.txt @@ -0,0 +1,11 @@ +set(K_WAY_HEAP "" CACHE STRING "K_WAY_HEAP") + +if(K_WAY_HEAP) + if(K_WAY_HEAP LESS 2) + message(FATAL_ERROR "K_WAY_HEAP value should be at least 2") + else() + set(CMAKE_CXX_SIM_FLAGS "-DK_WAY_HEAP=${K_WAY_HEAP}") + endif() +endif() + +add_subdirectory(src) diff --git a/src/dmclock/sim/dmc_sim_100th.conf b/src/dmclock/sim/dmc_sim_100th.conf new file mode 100644 index 000000000..17d004354 --- /dev/null +++ b/src/dmclock/sim/dmc_sim_100th.conf @@ -0,0 +1,32 @@ +[global] +server_groups = 1 +client_groups = 2 +server_random_selection = true +server_soft_limit = true + +[client.0] +client_count = 99 +client_wait = 0 +client_total_ops = 1000 +client_server_select_range = 10 +client_iops_goal = 50 +client_outstanding_ops = 100 +client_reservation = 20.0 +client_limit = 60.0 +client_weight = 1.0 + +[client.1] +client_count = 1 +client_wait = 10 +client_total_ops = 1000 +client_server_select_range = 10 +client_iops_goal = 50 +client_outstanding_ops = 100 +client_reservation = 20.0 +client_limit = 60.0 +client_weight = 1.0 + +[server.0] +server_count = 100 +server_iops = 40 +server_threads = 1 diff --git a/src/dmclock/sim/dmc_sim_example.conf b/src/dmclock/sim/dmc_sim_example.conf new file mode 100644 index 000000000..e98b870eb --- /dev/null +++ b/src/dmclock/sim/dmc_sim_example.conf @@ -0,0 +1,56 @@ +[global] +server_groups = 1 +client_groups = 4 +server_random_selection = false +server_soft_limit = false + +[client.0] +client_count = 1 +client_wait = 0 +client_total_ops = 2000 +client_server_select_range = 1 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 0.0 +client_limit = 0.0 +client_weight = 1.0 + +[client.1] +client_count = 1 +client_wait = 5 +client_total_ops = 2000 +client_server_select_range = 1 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 0.0 +client_limit = 40.0 +client_weight = 1.0 + +[client.2] +client_count = 1 +client_wait = 10 +client_total_ops = 2000 +client_server_select_range = 1 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 0.0 +client_limit = 50.0 +client_weight = 2.0 +client_req_cost = 1 + +[client.3] +client_count = 1 +client_wait = 10 +client_total_ops = 2000 +client_server_select_range = 1 +client_iops_goal = 200 +client_outstanding_ops = 32 +client_reservation = 0.0 +client_limit = 50.0 +client_weight = 2.0 +client_req_cost = 3 + +[server.0] +server_count = 1 +server_iops = 160 +server_threads = 1 diff --git a/src/dmclock/sim/src/CMakeLists.txt b/src/dmclock/sim/src/CMakeLists.txt new file mode 100644 index 000000000..dab011ed0 --- /dev/null +++ b/src/dmclock/sim/src/CMakeLists.txt @@ -0,0 +1,38 @@ +set(local_flags "-Wall ${CMAKE_CXX_SIM_FLAGS}") + +set(ssched_sim_srcs test_ssched.cc test_ssched_main.cc) +set(dmc_sim_srcs test_dmclock.cc test_dmclock_main.cc) +set(config_srcs config.cc str_list.cc ConfUtils.cc) + +set_source_files_properties(${ssched_sim_srcs} ${dmc_sim_srcs} ${dmc_srcs} ${config_srcs} + PROPERTIES + COMPILE_FLAGS "${local_flags}" + ) + +if ("${CMAKE_CXX_COMPILER_ID}" MATCHES "Clang") + set(warnings_off " -Wno-unused-variable -Wno-unused-function") +elseif ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU") + set(warnings_off " -Wno-unused-but-set-variable -Wno-unused-function") +endif() + +# append warning flags to certain source files +set_property( + SOURCE ${ssched_sim_srcs} ${dmc_sim_srcs} ${config_srcs} + APPEND_STRING + PROPERTY COMPILE_FLAGS "${warnings_off}" + ) + +add_executable(ssched_sim EXCLUDE_FROM_ALL ${ssched_sim_srcs}) +target_include_directories(ssched_sim PRIVATE ssched) # ssched code +add_executable(dmc_sim EXCLUDE_FROM_ALL ${dmc_sim_srcs} ${config_srcs}) + +set_target_properties(ssched_sim dmc_sim + PROPERTIES + RUNTIME_OUTPUT_DIRECTORY ..) + +add_dependencies(dmc_sim dmclock) + +target_link_libraries(ssched_sim LINK_PRIVATE Threads::Threads) +target_link_libraries(dmc_sim LINK_PRIVATE dmclock) + +add_custom_target(dmclock-sims DEPENDS ssched_sim dmc_sim) diff --git a/src/dmclock/sim/src/ConfUtils.cc b/src/dmclock/sim/src/ConfUtils.cc new file mode 100644 index 000000000..a05f7dc42 --- /dev/null +++ b/src/dmclock/sim/src/ConfUtils.cc @@ -0,0 +1,574 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab +/* + * Ceph - scalable distributed file system + * + * Copyright (C) 2011 New Dream Network + * + * This is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License version 2.1, as published by the Free Software + * Foundation. See file COPYING. + * + */ + +#include <algorithm> +#include <errno.h> +#include <list> +#include <map> +#include <sstream> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <string> +#include <sys/stat.h> +#include <sys/types.h> +#include <unistd.h> +#include <iostream> + +#include <assert.h> +#include "ConfUtils.h" + +using std::cerr; +using std::ostringstream; +using std::pair; +using std::string; + +#define MAX_CONFIG_FILE_SZ 0x40000000 + +////////////////////////////// ConfLine ////////////////////////////// +ConfLine:: +ConfLine(const std::string &key_, const std::string &val_, + const std::string &newsection_, const std::string &comment_, int line_no_) + : key(key_), val(val_), newsection(newsection_) +{ + // If you want to implement writable ConfFile support, you'll need to save + // the comment and line_no arguments here. +} + +bool ConfLine:: +operator<(const ConfLine &rhs) const +{ + // We only compare keys. + // If you have more than one line with the same key in a given section, the + // last one wins. + if (key < rhs.key) + return true; + else + return false; +} + +std::ostream &operator<<(std::ostream& oss, const ConfLine &l) +{ + oss << "ConfLine(key = '" << l.key << "', val='" + << l.val << "', newsection='" << l.newsection << "')"; + return oss; +} +///////////////////////// ConfFile ////////////////////////// +ConfFile:: +ConfFile() +{ +} + +ConfFile:: +~ConfFile() +{ +} + +void ConfFile:: +clear() +{ + sections.clear(); +} + +/* We load the whole file into memory and then parse it. Although this is not + * the optimal approach, it does mean that most of this code can be shared with + * the bufferlist loading function. Since bufferlists are always in-memory, the + * load_from_buffer interface works well for them. + * In general, configuration files should be a few kilobytes at maximum, so + * loading the whole configuration into memory shouldn't be a problem. + */ +int ConfFile:: +parse_file(const std::string &fname, std::deque<std::string> *errors, + std::ostream *warnings) +{ + clear(); + + int ret = 0; + size_t sz; + char *buf = NULL; + char buf2[128]; + FILE *fp = fopen(fname.c_str(), "r"); + if (!fp) { + ret = -errno; + return ret; + } + + struct stat st_buf; + if (fstat(fileno(fp), &st_buf)) { + ret = -errno; + ostringstream oss; + oss << "read_conf: failed to fstat '" << fname << "': " << strerror_r(ret, buf2, sizeof(buf2)); + errors->push_back(oss.str()); + goto done; + } + + if (st_buf.st_size > MAX_CONFIG_FILE_SZ) { + ostringstream oss; + oss << "read_conf: config file '" << fname << "' is " << st_buf.st_size + << " bytes, but the maximum is " << MAX_CONFIG_FILE_SZ; + errors->push_back(oss.str()); + ret = -EINVAL; + goto done; + } + + sz = (size_t)st_buf.st_size; + buf = (char*)malloc(sz); + if (!buf) { + ret = -ENOMEM; + goto done; + } + + if (fread(buf, 1, sz, fp) != sz) { + if (ferror(fp)) { + ret = -errno; + ostringstream oss; + oss << "read_conf: fread error while reading '" << fname << "': " + << strerror_r(ret, buf2, sizeof(buf2)); + errors->push_back(oss.str()); + goto done; + } + else { + ostringstream oss; + oss << "read_conf: unexpected EOF while reading '" << fname << "': " + << "possible concurrent modification?"; + errors->push_back(oss.str()); + ret = -EIO; + goto done; + } + } + + load_from_buffer(buf, sz, errors, warnings); + ret = 0; + +done: + free(buf); + fclose(fp); + return ret; +} + +int ConfFile:: +read(const std::string §ion, const std::string &key, std::string &val) const +{ + string k(normalize_key_name(key)); + + const_section_iter_t s = sections.find(section); + if (s == sections.end()) + return -ENOENT; + ConfLine exemplar(k, "", "", "", 0); + ConfSection::const_line_iter_t l = s->second.lines.find(exemplar); + if (l == s->second.lines.end()) + return -ENOENT; + val = l->val; + return 0; +} + +ConfFile::const_section_iter_t ConfFile:: +sections_begin() const +{ + return sections.begin(); +} + +ConfFile::const_section_iter_t ConfFile:: +sections_end() const +{ + return sections.end(); +} + +void ConfFile:: +trim_whitespace(std::string &str, bool strip_internal) +{ + // strip preceding + const char *in = str.c_str(); + while (true) { + char c = *in; + if ((!c) || (!isspace(c))) + break; + ++in; + } + char output[strlen(in) + 1]; + strcpy(output, in); + + // strip trailing + char *o = output + strlen(output); + while (true) { + if (o == output) + break; + --o; + if (!isspace(*o)) { + ++o; + *o = '\0'; + break; + } + } + + if (!strip_internal) { + str.assign(output); + return; + } + + // strip internal + char output2[strlen(output) + 1]; + char *out2 = output2; + bool prev_was_space = false; + for (char *u = output; *u; ++u) { + char c = *u; + if (isspace(c)) { + if (!prev_was_space) + *out2++ = c; + prev_was_space = true; + } + else { + *out2++ = c; + prev_was_space = false; + } + } + *out2++ = '\0'; + str.assign(output2); +} + +/* Normalize a key name. + * + * Normalized key names have no leading or trailing whitespace, and all + * whitespace is stored as underscores. The main reason for selecting this + * normal form is so that in common/config.cc, we can use a macro to stringify + * the field names of md_config_t and get a key in normal form. + */ +std::string ConfFile:: +normalize_key_name(const std::string &key) +{ + string k(key); + ConfFile::trim_whitespace(k, true); + std::replace(k.begin(), k.end(), ' ', '_'); + return k; +} + +std::ostream &operator<<(std::ostream &oss, const ConfFile &cf) +{ + for (ConfFile::const_section_iter_t s = cf.sections_begin(); + s != cf.sections_end(); ++s) { + oss << "[" << s->first << "]\n"; + for (ConfSection::const_line_iter_t l = s->second.lines.begin(); + l != s->second.lines.end(); ++l) { + if (!l->key.empty()) { + oss << "\t" << l->key << " = \"" << l->val << "\"\n"; + } + } + } + return oss; +} + +void ConfFile:: +load_from_buffer(const char *buf, size_t sz, std::deque<std::string> *errors, + std::ostream *warnings) +{ + errors->clear(); + + section_iter_t::value_type vt("global", ConfSection()); + pair < section_iter_t, bool > vr(sections.insert(vt)); + assert(vr.second); + section_iter_t cur_section = vr.first; + std::string acc; + + const char *b = buf; + int line_no = 0; + size_t line_len = -1; + size_t rem = sz; + while (1) { + b += line_len + 1; + rem -= line_len + 1; + if (rem == 0) + break; + line_no++; + + // look for the next newline + const char *end = (const char*)memchr(b, '\n', rem); + if (!end) { + ostringstream oss; + oss << "read_conf: ignoring line " << line_no << " because it doesn't " + << "end with a newline! Please end the config file with a newline."; + errors->push_back(oss.str()); + break; + } + + // find length of line, and search for NULLs + line_len = 0; + bool found_null = false; + for (const char *tmp = b; tmp != end; ++tmp) { + line_len++; + if (*tmp == '\0') { + found_null = true; + } + } + + if (found_null) { + ostringstream oss; + oss << "read_conf: ignoring line " << line_no << " because it has " + << "an embedded null."; + errors->push_back(oss.str()); + acc.clear(); + continue; + } + + if ((line_len >= 1) && (b[line_len-1] == '\\')) { + // A backslash at the end of a line serves as a line continuation marker. + // Combine the next line with this one. + // Remove the backslash itself from the text. + acc.append(b, line_len - 1); + continue; + } + + acc.append(b, line_len); + + //cerr << "acc = '" << acc << "'" << std::endl; + ConfLine *cline = process_line(line_no, acc.c_str(), errors); + acc.clear(); + if (!cline) + continue; + const std::string &csection(cline->newsection); + if (!csection.empty()) { + std::map <std::string, ConfSection>::value_type nt(csection, ConfSection()); + pair < section_iter_t, bool > nr(sections.insert(nt)); + cur_section = nr.first; + } + else { + if (cur_section->second.lines.count(*cline)) { + // replace an existing key/line in this section, so that + // [mysection] + // foo = 1 + // foo = 2 + // will result in foo = 2. + cur_section->second.lines.erase(*cline); + if (cline->key.length() && warnings) + *warnings << "warning: line " << line_no << ": '" << cline->key << "' in section '" + << cur_section->first << "' redefined " << std::endl; + } + // add line to current section + //std::cerr << "cur_section = " << cur_section->first << ", " << *cline << std::endl; + cur_section->second.lines.insert(*cline); + } + delete cline; + } + + if (!acc.empty()) { + ostringstream oss; + oss << "read_conf: don't end with lines that end in backslashes!"; + errors->push_back(oss.str()); + } +} + +/* + * A simple state-machine based parser. + * This probably could/should be rewritten with something like boost::spirit + * or yacc if the grammar ever gets more complex. + */ +ConfLine* ConfFile:: +process_line(int line_no, const char *line, std::deque<std::string> *errors) +{ + enum acceptor_state_t { + ACCEPT_INIT, + ACCEPT_SECTION_NAME, + ACCEPT_KEY, + ACCEPT_VAL_START, + ACCEPT_UNQUOTED_VAL, + ACCEPT_QUOTED_VAL, + ACCEPT_COMMENT_START, + ACCEPT_COMMENT_TEXT, + }; + const char *l = line; + acceptor_state_t state = ACCEPT_INIT; + string key, val, newsection, comment; + bool escaping = false; + while (true) { + char c = *l++; + switch (state) { + case ACCEPT_INIT: + if (c == '\0') + return NULL; // blank line. Not an error, but not interesting either. + else if (c == '[') + state = ACCEPT_SECTION_NAME; + else if ((c == '#') || (c == ';')) + state = ACCEPT_COMMENT_TEXT; + else if (c == ']') { + ostringstream oss; + oss << "unexpected right bracket at char " << (l - line) + << ", line " << line_no; + errors->push_back(oss.str()); + return NULL; + } + else if (isspace(c)) { + // ignore whitespace here + } + else { + // try to accept this character as a key + state = ACCEPT_KEY; + --l; + } + break; + case ACCEPT_SECTION_NAME: + if (c == '\0') { + ostringstream oss; + oss << "error parsing new section name: expected right bracket " + << "at char " << (l - line) << ", line " << line_no; + errors->push_back(oss.str()); + return NULL; + } + else if ((c == ']') && (!escaping)) { + trim_whitespace(newsection, true); + if (newsection.empty()) { + ostringstream oss; + oss << "error parsing new section name: no section name found? " + << "at char " << (l - line) << ", line " << line_no; + errors->push_back(oss.str()); + return NULL; + } + state = ACCEPT_COMMENT_START; + } + else if (((c == '#') || (c == ';')) && (!escaping)) { + ostringstream oss; + oss << "unexpected comment marker while parsing new section name, at " + << "char " << (l - line) << ", line " << line_no; + errors->push_back(oss.str()); + return NULL; + } + else if ((c == '\\') && (!escaping)) { + escaping = true; + } + else { + escaping = false; + newsection += c; + } + break; + case ACCEPT_KEY: + if ((((c == '#') || (c == ';')) && (!escaping)) || (c == '\0')) { + ostringstream oss; + if (c == '\0') { + oss << "end of key=val line " << line_no + << " reached, no \"=val\" found...missing =?"; + } else { + oss << "unexpected character while parsing putative key value, " + << "at char " << (l - line) << ", line " << line_no; + } + errors->push_back(oss.str()); + return NULL; + } + else if ((c == '=') && (!escaping)) { + key = normalize_key_name(key); + if (key.empty()) { + ostringstream oss; + oss << "error parsing key name: no key name found? " + << "at char " << (l - line) << ", line " << line_no; + errors->push_back(oss.str()); + return NULL; + } + state = ACCEPT_VAL_START; + } + else if ((c == '\\') && (!escaping)) { + escaping = true; + } + else { + escaping = false; + key += c; + } + break; + case ACCEPT_VAL_START: + if (c == '\0') + return new ConfLine(key, val, newsection, comment, line_no); + else if ((c == '#') || (c == ';')) + state = ACCEPT_COMMENT_TEXT; + else if (c == '"') + state = ACCEPT_QUOTED_VAL; + else if (isspace(c)) { + // ignore whitespace + } + else { + // try to accept character as a val + state = ACCEPT_UNQUOTED_VAL; + --l; + } + break; + case ACCEPT_UNQUOTED_VAL: + if (c == '\0') { + if (escaping) { + ostringstream oss; + oss << "error parsing value name: unterminated escape sequence " + << "at char " << (l - line) << ", line " << line_no; + errors->push_back(oss.str()); + return NULL; + } + trim_whitespace(val, false); + return new ConfLine(key, val, newsection, comment, line_no); + } + else if (((c == '#') || (c == ';')) && (!escaping)) { + trim_whitespace(val, false); + state = ACCEPT_COMMENT_TEXT; + } + else if ((c == '\\') && (!escaping)) { + escaping = true; + } + else { + escaping = false; + val += c; + } + break; + case ACCEPT_QUOTED_VAL: + if (c == '\0') { + ostringstream oss; + oss << "found opening quote for value, but not the closing quote. " + << "line " << line_no; + errors->push_back(oss.str()); + return NULL; + } + else if ((c == '"') && (!escaping)) { + state = ACCEPT_COMMENT_START; + } + else if ((c == '\\') && (!escaping)) { + escaping = true; + } + else { + escaping = false; + // Add anything, including whitespace. + val += c; + } + break; + case ACCEPT_COMMENT_START: + if (c == '\0') { + return new ConfLine(key, val, newsection, comment, line_no); + } + else if ((c == '#') || (c == ';')) { + state = ACCEPT_COMMENT_TEXT; + } + else if (isspace(c)) { + // ignore whitespace + } + else { + ostringstream oss; + oss << "unexpected character at char " << (l - line) << " of line " + << line_no; + errors->push_back(oss.str()); + return NULL; + } + break; + case ACCEPT_COMMENT_TEXT: + if (c == '\0') + return new ConfLine(key, val, newsection, comment, line_no); + else + comment += c; + break; + default: + assert(0); + break; + } + assert(c != '\0'); // We better not go past the end of the input string. + } +} diff --git a/src/dmclock/sim/src/ConfUtils.h b/src/dmclock/sim/src/ConfUtils.h new file mode 100644 index 000000000..3db1d1e14 --- /dev/null +++ b/src/dmclock/sim/src/ConfUtils.h @@ -0,0 +1,83 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab +/* + * Ceph - scalable distributed file system + * + * Copyright (C) 2011 New Dream Network + * + * This is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License version 2.1, as published by the Free Software + * Foundation. See file COPYING. + * + */ + +#ifndef CEPH_CONFUTILS_H +#define CEPH_CONFUTILS_H + +#include <deque> +#include <map> +#include <set> +#include <string> + +/* + * Ceph configuration file support. + * + * This class loads an INI-style configuration from a file or bufferlist, and + * holds it in memory. In general, an INI configuration file is composed of + * sections, which contain key/value pairs. You can put comments on the end of + * lines by using either a hash mark (#) or the semicolon (;). + * + * You can get information out of ConfFile by calling get_key or by examining + * individual sections. + * + * This class could be extended to support modifying configuration files and + * writing them back out without too much difficulty. Currently, this is not + * implemented, and the file is read-only. + */ +class ConfLine { +public: + ConfLine(const std::string &key_, const std::string &val_, + const std::string &newsection_, const std::string &comment_, int line_no_); + bool operator<(const ConfLine &rhs) const; + friend std::ostream &operator<<(std::ostream& oss, const ConfLine &l); + + std::string key, val, newsection; +}; + +class ConfSection { +public: + typedef std::set <ConfLine>::const_iterator const_line_iter_t; + + std::set <ConfLine> lines; +}; + +class ConfFile { +public: + typedef std::map <std::string, ConfSection>::iterator section_iter_t; + typedef std::map <std::string, ConfSection>::const_iterator const_section_iter_t; + + ConfFile(); + ~ConfFile(); + void clear(); + int parse_file(const std::string &fname, std::deque<std::string> *errors, std::ostream *warnings); + int read(const std::string §ion, const std::string &key, + std::string &val) const; + + const_section_iter_t sections_begin() const; + const_section_iter_t sections_end() const; + + static void trim_whitespace(std::string &str, bool strip_internal); + static std::string normalize_key_name(const std::string &key); + friend std::ostream &operator<<(std::ostream &oss, const ConfFile &cf); + +private: + void load_from_buffer(const char *buf, size_t sz, + std::deque<std::string> *errors, std::ostream *warnings); + static ConfLine* process_line(int line_no, const char *line, + std::deque<std::string> *errors); + + std::map <std::string, ConfSection> sections; +}; + +#endif diff --git a/src/dmclock/sim/src/config.cc b/src/dmclock/sim/src/config.cc new file mode 100644 index 000000000..dae2903e1 --- /dev/null +++ b/src/dmclock/sim/src/config.cc @@ -0,0 +1,184 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include <unistd.h> +#include <string.h> +#include <stdarg.h> + +#include <iostream> +#include <vector> +#include <list> + +#include "config.h" +#include "str_list.h" + + +static void dashes_to_underscores(const char *input, char *output) { + char c = 0; + char *o = output; + const char *i = input; + // first two characters are copied as-is + *o = *i++; + if (*o++ == '\0') + return; + *o = *i++; + if (*o++ == '\0') + return; + for (; ((c = *i)); ++i) { + if (c == '=') { + strcpy(o, i); + return; + } + if (c == '-') + *o++ = '_'; + else + *o++ = c; + } + *o++ = '\0'; +} + +static int va_ceph_argparse_witharg(std::vector<const char*> &args, + std::vector<const char*>::iterator &i, std::string *ret, + std::ostream &oss, va_list ap) { + const char *first = *i; + char tmp[strlen(first)+1]; + dashes_to_underscores(first, tmp); + first = tmp; + + // does this argument match any of the possibilities? + while (1) { + const char *a = va_arg(ap, char*); + if (a == NULL) + return 0; + int strlen_a = strlen(a); + char a2[strlen_a+1]; + dashes_to_underscores(a, a2); + if (strncmp(a2, first, strlen(a2)) == 0) { + if (first[strlen_a] == '=') { + *ret = first + strlen_a + 1; + i = args.erase(i); + return 1; + } + else if (first[strlen_a] == '\0') { + // find second part (or not) + if (i+1 == args.end()) { + oss << "Option " << *i << " requires an argument." << std::endl; + i = args.erase(i); + return -EINVAL; + } + i = args.erase(i); + *ret = *i; + i = args.erase(i); + return 1; + } + } + } +} + +bool crimson::qos_simulation::ceph_argparse_witharg(std::vector<const char*> &args, + std::vector<const char*>::iterator &i, std::string *ret, ...) { + int r; + va_list ap; + va_start(ap, ret); + r = va_ceph_argparse_witharg(args, i, ret, std::cerr, ap); + va_end(ap); + if (r < 0) + _exit(1); + return r != 0; +} + +void crimson::qos_simulation::ceph_argparse_early_args(std::vector<const char*>& args, std::string *conf_file_list) { + std::string val; + + std::vector<const char *> orig_args = args; + + for (std::vector<const char*>::iterator i = args.begin(); i != args.end(); ) { + if (ceph_argparse_witharg(args, i, &val, "--conf", "-c", (char*)NULL)) { + *conf_file_list = val; + } + else { + // ignore + ++i; + } + } + return; +} + +static bool stobool(const std::string & v) { + return !v.empty () && + (strcasecmp (v.c_str (), "true") == 0 || + atoi (v.c_str ()) != 0); +} + +int crimson::qos_simulation::parse_config_file(const std::string &fname, sim_config_t &g_conf) { + ConfFile cf; + std::deque<std::string> err; + std::ostringstream warn; + int ret = cf.parse_file(fname.c_str(), &err, &warn); + if (ret) { + // error + return ret; + } + + std::string val; + if (!cf.read("global", "server_groups", val)) + g_conf.server_groups = std::stoul(val); + if (!cf.read("global", "client_groups", val)) + g_conf.client_groups = std::stoul(val); + if (!cf.read("global", "server_random_selection", val)) + g_conf.server_random_selection = stobool(val); + if (!cf.read("global", "server_soft_limit", val)) + g_conf.server_soft_limit = stobool(val); + if (!cf.read("global", "anticipation_timeout", val)) + g_conf.anticipation_timeout = stod(val); + + for (unsigned i = 0; i < g_conf.server_groups; i++) { + srv_group_t st; + std::string section = "server." + std::to_string(i); + if (!cf.read(section, "server_count", val)) + st.server_count = std::stoul(val); + if (!cf.read(section, "server_iops", val)) + st.server_iops = std::stoul(val); + if (!cf.read(section, "server_threads", val)) + st.server_threads = std::stoul(val); + g_conf.srv_group.push_back(st); + } + + for (unsigned i = 0; i < g_conf.client_groups; i++) { + cli_group_t ct; + std::string section = "client." + std::to_string(i); + if (!cf.read(section, "client_count", val)) + ct.client_count = std::stoul(val); + if (!cf.read(section, "client_wait", val)) + ct.client_wait = std::chrono::seconds(std::stoul(val)); + if (!cf.read(section, "client_total_ops", val)) + ct.client_total_ops = std::stoul(val); + if (!cf.read(section, "client_server_select_range", val)) + ct.client_server_select_range = std::stoul(val); + if (!cf.read(section, "client_iops_goal", val)) + ct.client_iops_goal = std::stoul(val); + if (!cf.read(section, "client_outstanding_ops", val)) + ct.client_outstanding_ops = std::stoul(val); + if (!cf.read(section, "client_reservation", val)) + ct.client_reservation = std::stod(val); + if (!cf.read(section, "client_limit", val)) + ct.client_limit = std::stod(val); + if (!cf.read(section, "client_weight", val)) + ct.client_weight = std::stod(val); + if (!cf.read(section, "client_req_cost", val)) + ct.client_req_cost = std::stoul(val); + g_conf.cli_group.push_back(ct); + } + + return 0; +} diff --git a/src/dmclock/sim/src/config.h b/src/dmclock/sim/src/config.h new file mode 100644 index 000000000..d61ca45a8 --- /dev/null +++ b/src/dmclock/sim/src/config.h @@ -0,0 +1,158 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <string.h> + +#include <chrono> +#include <vector> +#include <sstream> +#include <iomanip> + +#include "ConfUtils.h" + +#include "sim_recs.h" + + +namespace crimson { + namespace qos_simulation { + + struct cli_group_t { + unsigned client_count; + std::chrono::seconds client_wait; + unsigned client_total_ops; + unsigned client_server_select_range; + unsigned client_iops_goal; + unsigned client_outstanding_ops; + double client_reservation; + double client_limit; + double client_weight; + Cost client_req_cost; + + cli_group_t(unsigned _client_count = 100, + unsigned _client_wait = 0, + unsigned _client_total_ops = 1000, + unsigned _client_server_select_range = 10, + unsigned _client_iops_goal = 50, + unsigned _client_outstanding_ops = 100, + double _client_reservation = 20.0, + double _client_limit = 60.0, + double _client_weight = 1.0, + Cost _client_req_cost = 1u) : + client_count(_client_count), + client_wait(std::chrono::seconds(_client_wait)), + client_total_ops(_client_total_ops), + client_server_select_range(_client_server_select_range), + client_iops_goal(_client_iops_goal), + client_outstanding_ops(_client_outstanding_ops), + client_reservation(_client_reservation), + client_limit(_client_limit), + client_weight(_client_weight), + client_req_cost(_client_req_cost) + { + // empty + } + + friend std::ostream& operator<<(std::ostream& out, + const cli_group_t& cli_group) { + out << + "client_count = " << cli_group.client_count << "\n" << + "client_wait = " << cli_group.client_wait.count() << "\n" << + "client_total_ops = " << cli_group.client_total_ops << "\n" << + "client_server_select_range = " << cli_group.client_server_select_range << "\n" << + "client_iops_goal = " << cli_group.client_iops_goal << "\n" << + "client_outstanding_ops = " << cli_group.client_outstanding_ops << "\n" << + std::fixed << std::setprecision(1) << + "client_reservation = " << cli_group.client_reservation << "\n" << + "client_limit = " << cli_group.client_limit << "\n" << + "client_weight = " << cli_group.client_weight << "\n" << + "client_req_cost = " << cli_group.client_req_cost; + return out; + } + }; // class cli_group_t + + + struct srv_group_t { + unsigned server_count; + unsigned server_iops; + unsigned server_threads; + + srv_group_t(unsigned _server_count = 100, + unsigned _server_iops = 40, + unsigned _server_threads = 1) : + server_count(_server_count), + server_iops(_server_iops), + server_threads(_server_threads) + { + // empty + } + + friend std::ostream& operator<<(std::ostream& out, + const srv_group_t& srv_group) { + out << + "server_count = " << srv_group.server_count << "\n" << + "server_iops = " << srv_group.server_iops << "\n" << + "server_threads = " << srv_group.server_threads; + return out; + } + }; // class srv_group_t + + + struct sim_config_t { + unsigned server_groups; + unsigned client_groups; + bool server_random_selection; + bool server_soft_limit; + double anticipation_timeout; + + std::vector<cli_group_t> cli_group; + std::vector<srv_group_t> srv_group; + + sim_config_t(unsigned _server_groups = 1, + unsigned _client_groups = 1, + bool _server_random_selection = false, + bool _server_soft_limit = true, + double _anticipation_timeout = 0.0) : + server_groups(_server_groups), + client_groups(_client_groups), + server_random_selection(_server_random_selection), + server_soft_limit(_server_soft_limit), + anticipation_timeout(_anticipation_timeout) + { + srv_group.reserve(server_groups); + cli_group.reserve(client_groups); + } + + friend std::ostream& operator<<(std::ostream& out, + const sim_config_t& sim_config) { + out << + "server_groups = " << sim_config.server_groups << "\n" << + "client_groups = " << sim_config.client_groups << "\n" << + "server_random_selection = " << sim_config.server_random_selection << "\n" << + "server_soft_limit = " << sim_config.server_soft_limit << "\n" << + std::fixed << std::setprecision(3) << + "anticipation_timeout = " << sim_config.anticipation_timeout; + return out; + } + }; // class sim_config_t + + + bool ceph_argparse_witharg(std::vector<const char*> &args, + std::vector<const char*>::iterator &i, std::string *ret, ...); + void ceph_argparse_early_args(std::vector<const char*>& args, std::string *conf_file_list); + int parse_config_file(const std::string &fname, sim_config_t &g_conf); + + }; // namespace qos_simulation +}; // namespace crimson diff --git a/src/dmclock/sim/src/sim_client.h b/src/dmclock/sim/src/sim_client.h new file mode 100644 index 000000000..182cdb803 --- /dev/null +++ b/src/dmclock/sim/src/sim_client.h @@ -0,0 +1,340 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <atomic> +#include <mutex> +#include <condition_variable> +#include <thread> +#include <chrono> +#include <vector> +#include <deque> +#include <iostream> + +#include "sim_recs.h" + + +namespace crimson { + namespace qos_simulation { + + struct req_op_t {}; + struct wait_op_t {}; + constexpr struct req_op_t req_op {}; + constexpr struct wait_op_t wait_op {}; + + + enum class CliOp { req, wait }; + struct CliInst { + CliOp op; + union { + std::chrono::milliseconds wait_time; + struct { + uint32_t count; + std::chrono::microseconds time_bw_reqs; + uint16_t max_outstanding; + } req_params; + } args; + + // D is a duration type + template<typename D> + CliInst(wait_op_t, D duration) : + op(CliOp::wait) + { + args.wait_time = + std::chrono::duration_cast<std::chrono::milliseconds>(duration); + } + + CliInst(req_op_t, + uint32_t count, double ops_per_sec, uint16_t max_outstanding) : + op(CliOp::req) + { + args.req_params.count = count; + args.req_params.max_outstanding = max_outstanding; + uint32_t us = uint32_t(0.5 + 1.0 / ops_per_sec * 1000000); + args.req_params.time_bw_reqs = std::chrono::microseconds(us); + } + }; + + + using ServerSelectFunc = std::function<const ServerId&(uint64_t seed)>; + + + template<typename SvcTrk, typename ReqPm, typename RespPm, typename Accum> + class SimulatedClient { + public: + + struct InternalStats { + std::mutex mtx; + std::chrono::nanoseconds track_resp_time; + std::chrono::nanoseconds get_req_params_time; + uint32_t track_resp_count; + uint32_t get_req_params_count; + + InternalStats() : + track_resp_time(0), + get_req_params_time(0), + track_resp_count(0), + get_req_params_count(0) + { + // empty + } + }; + + using SubmitFunc = + std::function<void(const ServerId&, + TestRequest&&, + const ClientId&, + const ReqPm&)>; + + using ClientAccumFunc = std::function<void(Accum&,const RespPm&)>; + + typedef std::chrono::time_point<std::chrono::steady_clock> TimePoint; + + static TimePoint now() { return std::chrono::steady_clock::now(); } + + protected: + + struct RespQueueItem { + TestResponse response; + ServerId server_id; + RespPm resp_params; + Cost request_cost; + }; + + const ClientId id; + const SubmitFunc submit_f; + const ServerSelectFunc server_select_f; + const ClientAccumFunc accum_f; + + std::vector<CliInst> instructions; + + SvcTrk service_tracker; + + // TODO: use lock rather than atomic??? + std::atomic_ulong outstanding_ops; + std::atomic_bool requests_complete; + + std::deque<RespQueueItem> resp_queue; + + std::mutex mtx_req; + std::condition_variable cv_req; + + std::mutex mtx_resp; + std::condition_variable cv_resp; + + using RespGuard = std::lock_guard<decltype(mtx_resp)>; + using Lock = std::unique_lock<std::mutex>; + + // data collection + + std::vector<TimePoint> op_times; + Accum accumulator; + InternalStats internal_stats; + + std::thread thd_req; + std::thread thd_resp; + + public: + + SimulatedClient(ClientId _id, + const SubmitFunc& _submit_f, + const ServerSelectFunc& _server_select_f, + const ClientAccumFunc& _accum_f, + const std::vector<CliInst>& _instrs) : + id(_id), + submit_f(_submit_f), + server_select_f(_server_select_f), + accum_f(_accum_f), + instructions(_instrs), + service_tracker(), + outstanding_ops(0), + requests_complete(false) + { + size_t op_count = 0; + for (auto i : instructions) { + if (CliOp::req == i.op) { + op_count += i.args.req_params.count; + } + } + op_times.reserve(op_count); + + thd_resp = std::thread(&SimulatedClient::run_resp, this); + thd_req = std::thread(&SimulatedClient::run_req, this); + } + + + SimulatedClient(ClientId _id, + const SubmitFunc& _submit_f, + const ServerSelectFunc& _server_select_f, + const ClientAccumFunc& _accum_f, + uint16_t _ops_to_run, + double _iops_goal, + uint16_t _outstanding_ops_allowed) : + SimulatedClient(_id, + _submit_f, _server_select_f, _accum_f, + {{req_op, _ops_to_run, _iops_goal, _outstanding_ops_allowed}}) + { + // empty + } + + + SimulatedClient(const SimulatedClient&) = delete; + SimulatedClient(SimulatedClient&&) = delete; + SimulatedClient& operator=(const SimulatedClient&) = delete; + SimulatedClient& operator=(SimulatedClient&&) = delete; + + virtual ~SimulatedClient() { + wait_until_done(); + } + + void receive_response(const TestResponse& resp, + const ServerId& server_id, + const RespPm& resp_params, + const Cost request_cost) { + RespGuard g(mtx_resp); + resp_queue.push_back( + RespQueueItem{ resp, server_id, resp_params, request_cost }); + cv_resp.notify_one(); + } + + const std::vector<TimePoint>& get_op_times() const { return op_times; } + + void wait_until_done() { + if (thd_req.joinable()) thd_req.join(); + if (thd_resp.joinable()) thd_resp.join(); + } + + const Accum& get_accumulator() const { return accumulator; } + + const InternalStats& get_internal_stats() const { return internal_stats; } + + protected: + + void run_req() { + size_t ops_count = 0; + for (auto i : instructions) { + if (CliOp::wait == i.op) { + std::this_thread::sleep_for(i.args.wait_time); + } else if (CliOp::req == i.op) { + Lock l(mtx_req); + for (uint64_t o = 0; o < i.args.req_params.count; ++o) { + while (outstanding_ops >= i.args.req_params.max_outstanding) { + cv_req.wait(l); + } + + l.unlock(); + auto now = std::chrono::steady_clock::now(); + const ServerId& server = server_select_f(o); + + ReqPm rp = + time_stats_w_return<decltype(internal_stats.get_req_params_time), + ReqPm>(internal_stats.mtx, + internal_stats.get_req_params_time, + [&]() -> ReqPm { + return service_tracker.get_req_params(server); + }); + count_stats(internal_stats.mtx, + internal_stats.get_req_params_count); + + submit_f(server, + TestRequest{server, static_cast<uint32_t>(o), 12}, + id, rp); + ++outstanding_ops; + l.lock(); // lock for return to top of loop + + auto delay_time = now + i.args.req_params.time_bw_reqs; + while (std::chrono::steady_clock::now() < delay_time) { + cv_req.wait_until(l, delay_time); + } // while + } // for + ops_count += i.args.req_params.count; + } else { + assert(false); + } + } // for loop + + requests_complete = true; + + // all requests made, thread ends + } + + + void run_resp() { + std::chrono::milliseconds delay(1000); + int op = 0; + + Lock l(mtx_resp); + + // since the following code would otherwise be repeated (except for + // the call to notify_one) in the two loops below; let's avoid + // repetition and define it once. + const auto proc_resp = [this, &op, &l](const bool notify_req_cv) { + if (!resp_queue.empty()) { + RespQueueItem item = resp_queue.front(); + resp_queue.pop_front(); + + l.unlock(); + + // data collection + + op_times.push_back(now()); + accum_f(accumulator, item.resp_params); + + // processing + +#if 0 // not needed + TestResponse& resp = item.response; +#endif + + time_stats(internal_stats.mtx, + internal_stats.track_resp_time, + [&](){ + service_tracker.track_resp(item.server_id, item.resp_params, item.request_cost); + }); + count_stats(internal_stats.mtx, + internal_stats.track_resp_count); + + --outstanding_ops; + if (notify_req_cv) { + cv_req.notify_one(); + } + + l.lock(); + } + }; + + while(!requests_complete.load()) { + while(resp_queue.empty() && !requests_complete.load()) { + cv_resp.wait_for(l, delay); + } + proc_resp(true); + } + + while(outstanding_ops.load() > 0) { + while(resp_queue.empty() && outstanding_ops.load() > 0) { + cv_resp.wait_for(l, delay); + } + proc_resp(false); // don't call notify_one as all requests are complete + } + + // all responses received, thread ends + } + }; // class SimulatedClient + + + }; // namespace qos_simulation +}; // namespace crimson diff --git a/src/dmclock/sim/src/sim_recs.h b/src/dmclock/sim/src/sim_recs.h new file mode 100644 index 000000000..010630072 --- /dev/null +++ b/src/dmclock/sim/src/sim_recs.h @@ -0,0 +1,131 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <stdint.h> +#include <stdlib.h> +#include <assert.h> +#include <signal.h> + +#include <sys/time.h> + +#include <cmath> +#include <limits> +#include <string> +#include <mutex> +#include <iostream> +#include <functional> + + +using ClientId = unsigned; +using ServerId = unsigned; + + +namespace crimson { + namespace qos_simulation { + + using Cost = uint32_t; + + inline void debugger() { + raise(SIGCONT); + } + + template<typename T> + void time_stats(std::mutex& mtx, + T& time_accumulate, + std::function<void()> code) { + auto t1 = std::chrono::steady_clock::now(); + code(); + auto t2 = std::chrono::steady_clock::now(); + auto duration = t2 - t1; + auto cast_duration = std::chrono::duration_cast<T>(duration); + std::lock_guard<std::mutex> lock(mtx); + time_accumulate += cast_duration; + } + + // unfortunately it's hard for the compiler to infer the types, + // and therefore when called the template params might have to be + // explicit + template<typename T, typename R> + R time_stats_w_return(std::mutex& mtx, + T& time_accumulate, + std::function<R()> code) { + auto t1 = std::chrono::steady_clock::now(); + R result = code(); + auto t2 = std::chrono::steady_clock::now(); + auto duration = t2 - t1; + auto cast_duration = std::chrono::duration_cast<T>(duration); + std::lock_guard<std::mutex> lock(mtx); + time_accumulate += cast_duration; + return result; + } + + template<typename T> + void count_stats(std::mutex& mtx, + T& counter) { + std::lock_guard<std::mutex> lock(mtx); + ++counter; + } + + struct TestRequest { + ServerId server; // allows debugging + uint32_t epoch; + uint32_t op; + + TestRequest(ServerId _server, + uint32_t _epoch, + uint32_t _op) : + server(_server), + epoch(_epoch), + op(_op) + { + // empty + } + + TestRequest(const TestRequest& r) : + TestRequest(r.server, r.epoch, r.op) + { + // empty + } + }; // struct TestRequest + + + struct TestResponse { + uint32_t epoch; + + explicit TestResponse(uint32_t _epoch) : + epoch(_epoch) + { + // empty + } + + TestResponse(const TestResponse& r) : + epoch(r.epoch) + { + // empty + } + + friend std::ostream& operator<<(std::ostream& out, const TestResponse& resp) { + out << "{ "; + out << "epoch:" << resp.epoch; + out << " }"; + return out; + } + }; // class TestResponse + + }; // namespace qos_simulation +}; // namespace crimson diff --git a/src/dmclock/sim/src/sim_server.h b/src/dmclock/sim/src/sim_server.h new file mode 100644 index 000000000..743c6e79a --- /dev/null +++ b/src/dmclock/sim/src/sim_server.h @@ -0,0 +1,245 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <thread> +#include <mutex> +#include <condition_variable> +#include <chrono> +#include <deque> + +#include "sim_recs.h" + + +namespace crimson { + namespace qos_simulation { + + template<typename Q, typename ReqPm, typename RespPm, typename Accum> + class SimulatedServer { + + struct QueueItem { + ClientId client; + std::unique_ptr<TestRequest> request; + RespPm additional; + Cost request_cost; + + QueueItem(const ClientId& _client, + std::unique_ptr<TestRequest>&& _request, + const RespPm& _additional, + const Cost _request_cost) : + client(_client), + request(std::move(_request)), + additional(_additional), + request_cost(_request_cost) + { + // empty + } + }; // QueueItem + + public: + + struct InternalStats { + std::mutex mtx; + std::chrono::nanoseconds add_request_time; + std::chrono::nanoseconds request_complete_time; + uint32_t add_request_count; + uint32_t request_complete_count; + + InternalStats() : + add_request_time(0), + request_complete_time(0), + add_request_count(0), + request_complete_count(0) + { + // empty + } + }; + + using ClientRespFunc = std::function<void(ClientId, + const TestResponse&, + const ServerId&, + const RespPm&, + const Cost)>; + + using ServerAccumFunc = std::function<void(Accum& accumulator, + const RespPm& additional)>; + + protected: + + const ServerId id; + Q* priority_queue; + ClientRespFunc client_resp_f; + int iops; + size_t thread_pool_size; + + bool finishing; + std::chrono::microseconds op_time; + + std::mutex inner_queue_mtx; + std::condition_variable inner_queue_cv; + std::deque<QueueItem> inner_queue; + + std::thread* threads; + + using InnerQGuard = std::lock_guard<decltype(inner_queue_mtx)>; + using Lock = std::unique_lock<std::mutex>; + + // data collection + + ServerAccumFunc accum_f; + Accum accumulator; + + InternalStats internal_stats; + + public: + + using CanHandleRequestFunc = std::function<bool(void)>; + using HandleRequestFunc = + std::function<void(const ClientId&,std::unique_ptr<TestRequest>,const RespPm&, uint64_t)>; + using CreateQueueF = std::function<Q*(CanHandleRequestFunc,HandleRequestFunc)>; + + + SimulatedServer(ServerId _id, + int _iops, + size_t _thread_pool_size, + const ClientRespFunc& _client_resp_f, + const ServerAccumFunc& _accum_f, + CreateQueueF _create_queue_f) : + id(_id), + priority_queue(_create_queue_f(std::bind(&SimulatedServer::has_avail_thread, + this), + std::bind(&SimulatedServer::inner_post, + this, + std::placeholders::_1, + std::placeholders::_2, + std::placeholders::_3, + std::placeholders::_4))), + client_resp_f(_client_resp_f), + iops(_iops), + thread_pool_size(_thread_pool_size), + finishing(false), + accum_f(_accum_f) + { + op_time = + std::chrono::microseconds((int) (0.5 + + thread_pool_size * 1000000.0 / iops)); + std::chrono::milliseconds finishing_check_period(1000); + threads = new std::thread[thread_pool_size]; + for (size_t i = 0; i < thread_pool_size; ++i) { + threads[i] = std::thread(&SimulatedServer::run, this, finishing_check_period); + } + } + + virtual ~SimulatedServer() { + Lock l(inner_queue_mtx); + finishing = true; + inner_queue_cv.notify_all(); + l.unlock(); + + for (size_t i = 0; i < thread_pool_size; ++i) { + threads[i].join(); + } + + delete[] threads; + + delete priority_queue; + } + + void post(TestRequest&& request, + const ClientId& client_id, + const ReqPm& req_params, + const Cost request_cost) + { + time_stats(internal_stats.mtx, + internal_stats.add_request_time, + [&](){ + priority_queue->add_request(std::move(request), + client_id, + req_params, + request_cost); + }); + count_stats(internal_stats.mtx, + internal_stats.add_request_count); + } + + bool has_avail_thread() { + InnerQGuard g(inner_queue_mtx); + return inner_queue.size() <= thread_pool_size; + } + + const Accum& get_accumulator() const { return accumulator; } + const Q& get_priority_queue() const { return *priority_queue; } + const InternalStats& get_internal_stats() const { return internal_stats; } + + protected: + + void inner_post(const ClientId& client, + std::unique_ptr<TestRequest> request, + const RespPm& additional, + const Cost request_cost) { + Lock l(inner_queue_mtx); + assert(!finishing); + accum_f(accumulator, additional); + inner_queue.emplace_back(QueueItem(client, + std::move(request), + additional, + request_cost)); + inner_queue_cv.notify_one(); + } + + void run(std::chrono::milliseconds check_period) { + Lock l(inner_queue_mtx); + while(true) { + while(inner_queue.empty() && !finishing) { + inner_queue_cv.wait_for(l, check_period); + } + if (!inner_queue.empty()) { + auto& front = inner_queue.front(); + auto client = front.client; + auto req = std::move(front.request); + auto additional = front.additional; + auto request_cost = front.request_cost; + inner_queue.pop_front(); + + l.unlock(); + + // simulation operation by sleeping; then call function to + // notify server of completion + std::this_thread::sleep_for(op_time * request_cost); + + // TODO: rather than assuming this constructor exists, perhaps + // pass in a function that does this mapping? + client_resp_f(client, TestResponse{req->epoch}, id, additional, request_cost); + + time_stats(internal_stats.mtx, + internal_stats.request_complete_time, + [&](){ + priority_queue->request_completed(); + }); + count_stats(internal_stats.mtx, + internal_stats.request_complete_count); + + l.lock(); // in prep for next iteration of loop + } else { + break; + } + } + } + }; // class SimulatedServer + + }; // namespace qos_simulation +}; // namespace crimson diff --git a/src/dmclock/sim/src/simulate.h b/src/dmclock/sim/src/simulate.h new file mode 100644 index 000000000..44a09ca31 --- /dev/null +++ b/src/dmclock/sim/src/simulate.h @@ -0,0 +1,448 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <assert.h> + +#include <memory> +#include <chrono> +#include <map> +#include <random> +#include <iostream> +#include <iomanip> +#include <string> + + +namespace crimson { + namespace qos_simulation { + + template<typename ServerId, typename ClientId, typename TS, typename TC> + class Simulation { + + public: + + using TimePoint = std::chrono::time_point<std::chrono::steady_clock>; + + protected: + + using ClientMap = std::map<ClientId,TC*>; + using ServerMap = std::map<ServerId,TS*>; + + unsigned server_count = 0; + unsigned client_count = 0; + + ServerMap servers; + ClientMap clients; + std::vector<ServerId> server_ids; + + TimePoint early_time; + TimePoint servers_created_time; + TimePoint clients_created_time; + TimePoint clients_finished_time; + TimePoint late_time; + + std::default_random_engine prng; + + bool has_run = false; + + + public: + + double fmt_tp(const TimePoint& t) { + auto c = t.time_since_epoch().count(); + return uint64_t(c / 1000000.0 + 0.5) % 100000 / 1000.0; + } + + TimePoint now() { + return std::chrono::steady_clock::now(); + } + + using ClientBasedServerSelectFunc = + std::function<const ServerId&(uint64_t, uint16_t)>; + + using ClientFilter = std::function<bool(const ClientId&)>; + + using ServerFilter = std::function<bool(const ServerId&)>; + + using ServerDataOutF = + std::function<void(std::ostream& out, + Simulation* sim, ServerFilter, + int header_w, int data_w, int data_prec)>; + + using ClientDataOutF = + std::function<void(std::ostream& out, + Simulation* sim, ClientFilter, + int header_w, int data_w, int data_prec)>; + + Simulation() : + early_time(now()), + prng(std::chrono::system_clock::now().time_since_epoch().count()) + { + // empty + } + + ~Simulation() { + for (auto c : clients) { + TC* cp = c.second; + delete cp; + } + + for (auto s : servers) { + delete s.second; + } + } + + unsigned get_client_count() const { return client_count; } + unsigned get_server_count() const { return server_count; } + TC& get_client(ClientId id) { return *clients[id]; } + TS& get_server(ServerId id) { return *servers[id]; } + const ServerId& get_server_id(std::size_t index) const { + return server_ids[index]; + } + + + void add_servers(unsigned count, + std::function<TS*(ServerId)> create_server_f) { + unsigned i = server_count; + + // increment server_count before creating servers since they + // will start running immediately and may use the server_count + // value; NB: this could still be an issue if servers are + // added with multiple add_servers calls; consider using a + // separate start function after all servers (and clients?) + // have been added + server_count += count; + + for (; i < server_count; ++i) { + server_ids.push_back(i); + servers[i] = create_server_f(i); + } + + servers_created_time = now(); + } + + + void add_clients(unsigned count, + std::function<TC*(ClientId)> create_client_f) { + unsigned i = client_count; + + // increment client_count before creating clients since they + // will start running immediately and may use the client_count + // value (e.g., in the server selection function); NB: this could + // still be an issue if clients are added with multiple + // add_clients calls; consider using a separate start function + // after all clients have been added + client_count += count; + + for (; i < client_count; ++i) { + clients[i] = create_client_f(i); + } + + clients_created_time = now(); + } + + + void run() { + assert(server_count > 0); + assert(client_count > 0); + + std::cout << "simulation started" << std::endl; + + // clients are now running; wait for all to finish + + for (auto const &i : clients) { + i.second->wait_until_done(); + } + + late_time = clients_finished_time = now(); + + std::cout << "simulation completed in " << + std::chrono::duration_cast<std::chrono::milliseconds>(clients_finished_time - servers_created_time).count() << + " millisecs" << std::endl; + + has_run = true; + } // run + + + void display_stats(std::ostream& out, + ServerDataOutF server_out_f, ClientDataOutF client_out_f, + ServerFilter server_filter = + [] (const ServerId&) { return true; }, + ClientFilter client_filter = + [] (const ClientId&) { return true; }, + int head_w = 12, int data_w = 7, int data_prec = 2) { + assert(has_run); + + // skip first 2 secondsd of data + const std::chrono::seconds skip_amount(0); + // calculate in groups of 5 seconds + const std::chrono::seconds measure_unit(2); + // unit to output reports in + const std::chrono::seconds report_unit(1); + + // compute and display stats + + TimePoint earliest_start = late_time; + TimePoint latest_start = early_time; + TimePoint earliest_finish = late_time; + TimePoint latest_finish = early_time; + + for (auto const &c : clients) { + auto start = c.second->get_op_times().front(); + auto end = c.second->get_op_times().back(); + + if (start < earliest_start) { earliest_start = start; } + if (start > latest_start) { latest_start = start; } + if (end < earliest_finish) { earliest_finish = end; } + if (end > latest_finish) { latest_finish = end; } + } + + double ops_factor = + std::chrono::duration_cast<std::chrono::duration<double>>(measure_unit) / + std::chrono::duration_cast<std::chrono::duration<double>>(report_unit); + + const auto start_edge = clients_created_time + skip_amount; + + std::map<ClientId,std::vector<double>> ops_data; + + for (auto const &c : clients) { + auto it = c.second->get_op_times().begin(); + const auto end = c.second->get_op_times().end(); + while (it != end && *it < start_edge) { ++it; } + + for (auto time_edge = start_edge + measure_unit; + time_edge <= latest_finish + measure_unit; + time_edge += measure_unit) { + int count = 0; + for (; it != end && *it < time_edge; ++count, ++it) { /* empty */ } + double ops_per_second = double(count) / ops_factor; + ops_data[c.first].push_back(ops_per_second); + } + } + + out << "==== Client Data ====" << std::endl; + + out << std::setw(head_w) << "client:"; + for (auto const &c : clients) { + if (!client_filter(c.first)) continue; + out << " " << std::setw(data_w) << c.first; + } + out << std::setw(data_w) << "total" << std::endl; + + { + bool has_data; + size_t i = 0; + do { + std::string line_header = "t_" + std::to_string(i) + ":"; + out << std::setw(head_w) << line_header; + has_data = false; + double total = 0.0; + for (auto const &c : clients) { + double data = 0.0; + if (i < ops_data[c.first].size()) { + data = ops_data[c.first][i]; + has_data = true; + } + total += data; + + if (!client_filter(c.first)) continue; + + out << " " << std::setw(data_w) << std::setprecision(data_prec) << + std::fixed << data; + } + out << " " << std::setw(data_w) << std::setprecision(data_prec) << + std::fixed << total << std::endl; + ++i; + } while(has_data); + } + + client_out_f(out, this, client_filter, head_w, data_w, data_prec); + + display_client_internal_stats<std::chrono::nanoseconds>(out, + "nanoseconds"); + + out << std::endl << "==== Server Data ====" << std::endl; + + out << std::setw(head_w) << "server:"; + for (auto const &s : servers) { + if (!server_filter(s.first)) continue; + out << " " << std::setw(data_w) << s.first; + } + out << " " << std::setw(data_w) << "total" << std::endl; + + server_out_f(out, this, server_filter, head_w, data_w, data_prec); + + display_server_internal_stats<std::chrono::nanoseconds>(out, + "nanoseconds"); + + // clean up clients then servers + + for (auto i = clients.begin(); i != clients.end(); ++i) { + delete i->second; + i->second = nullptr; + } + + for (auto i = servers.begin(); i != servers.end(); ++i) { + delete i->second; + i->second = nullptr; + } + } // display_stats + + + template<typename T> + void display_server_internal_stats(std::ostream& out, + const std::string& time_unit) { + T add_request_time(0); + T request_complete_time(0); + uint32_t add_request_count = 0; + uint32_t request_complete_count = 0; + + for (unsigned i = 0; i < get_server_count(); ++i) { + const auto& server = get_server(i); + const auto& is = server.get_internal_stats(); + add_request_time += + std::chrono::duration_cast<T>(is.add_request_time); + request_complete_time += + std::chrono::duration_cast<T>(is.request_complete_time); + add_request_count += is.add_request_count; + request_complete_count += is.request_complete_count; + } + + double add_request_time_per_unit = + double(add_request_time.count()) / add_request_count ; + out << "total time to add requests: " << + std::fixed << add_request_time.count() << " " << time_unit << + ";" << std::endl << + " count: " << add_request_count << ";" << std::endl << + " average: " << add_request_time_per_unit << + " " << time_unit << " per request/response" << std::endl; + + double request_complete_time_unit = + double(request_complete_time.count()) / request_complete_count ; + out << "total time to note requests complete: " << std::fixed << + request_complete_time.count() << " " << time_unit << ";" << + std::endl << + " count: " << request_complete_count << ";" << std::endl << + " average: " << request_complete_time_unit << + " " << time_unit << " per request/response" << std::endl; + + out << std::endl; + + assert(add_request_count == request_complete_count); + out << "server timing for QOS algorithm: " << + add_request_time_per_unit + request_complete_time_unit << + " " << time_unit << " per request/response" << std::endl; + } + + + template<typename T> + void display_client_internal_stats(std::ostream& out, + const std::string& time_unit) { + T track_resp_time(0); + T get_req_params_time(0); + uint32_t track_resp_count = 0; + uint32_t get_req_params_count = 0; + + for (unsigned i = 0; i < get_client_count(); ++i) { + const auto& client = get_client(i); + const auto& is = client.get_internal_stats(); + track_resp_time += + std::chrono::duration_cast<T>(is.track_resp_time); + get_req_params_time += + std::chrono::duration_cast<T>(is.get_req_params_time); + track_resp_count += is.track_resp_count; + get_req_params_count += is.get_req_params_count; + } + + double track_resp_time_unit = + double(track_resp_time.count()) / track_resp_count; + out << "total time to track responses: " << + std::fixed << track_resp_time.count() << " " << time_unit << ";" << + std::endl << + " count: " << track_resp_count << ";" << std::endl << + " average: " << track_resp_time_unit << " " << time_unit << + " per request/response" << std::endl; + + double get_req_params_time_unit = + double(get_req_params_time.count()) / get_req_params_count; + out << "total time to get request parameters: " << + std::fixed << get_req_params_time.count() << " " << time_unit << + ";" << std::endl << + " count: " << get_req_params_count << ";" << std::endl << + " average: " << get_req_params_time_unit << " " << time_unit << + " per request/response" << std::endl; + + out << std::endl; + + assert(track_resp_count == get_req_params_count); + out << "client timing for QOS algorithm: " << + track_resp_time_unit + get_req_params_time_unit << " " << + time_unit << " per request/response" << std::endl; + } + + + // **** server selection functions **** + + + const ServerId& server_select_alternate(uint64_t seed, + uint16_t client_idx) { + size_t index = (client_idx + seed) % server_count; + return server_ids[index]; + } + + + // returns a lambda using the range specified as servers_per (client) + ClientBasedServerSelectFunc + make_server_select_alt_range(uint16_t servers_per) { + return [servers_per,this](uint64_t seed, uint16_t client_idx) + -> const ServerId& { + double factor = double(server_count) / client_count; + size_t offset = seed % servers_per; + size_t index = (size_t(0.5 + client_idx * factor) + offset) % server_count; + return server_ids[index]; + }; + } + + + // function to choose a server randomly + const ServerId& server_select_random(uint64_t seed, uint16_t client_idx) { + size_t index = prng() % server_count; + return server_ids[index]; + } + + + // function to choose a server randomly + ClientBasedServerSelectFunc + make_server_select_ran_range(uint16_t servers_per) { + return [servers_per,this](uint64_t seed, uint16_t client_idx) + -> const ServerId& { + double factor = double(server_count) / client_count; + size_t offset = prng() % servers_per; + size_t index = (size_t(0.5 + client_idx * factor) + offset) % server_count; + return server_ids[index]; + }; + } + + + // function to always choose the first server + const ServerId& server_select_0(uint64_t seed, uint16_t client_idx) { + return server_ids[0]; + } + }; // class Simulation + + }; // namespace qos_simulation +}; // namespace crimson diff --git a/src/dmclock/sim/src/ssched/ssched_client.h b/src/dmclock/sim/src/ssched/ssched_client.h new file mode 100644 index 000000000..6764a09a8 --- /dev/null +++ b/src/dmclock/sim/src/ssched/ssched_client.h @@ -0,0 +1,51 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + +#include "ssched_recs.h" + + +namespace crimson { + namespace simple_scheduler { + + // S is server identifier type + template<typename S> + class ServiceTracker { + + public: + + // we have to start the counters at 1, as 0 is used in the + // cleaning process + ServiceTracker() + { + // empty + } + + void track_resp(const S& server_id, + const NullData& ignore, + uint64_t request_cost) { + // empty + } + + /* + * Returns the ReqParams for the given server. + */ + ReqParams get_req_params(const S& server) { + return ReqParams(); + } // get_req_params + }; // class ServiceTracker + } // namespace simple_scheduler +} // namespace crimson diff --git a/src/dmclock/sim/src/ssched/ssched_recs.h b/src/dmclock/sim/src/ssched/ssched_recs.h new file mode 100644 index 000000000..935e678c1 --- /dev/null +++ b/src/dmclock/sim/src/ssched/ssched_recs.h @@ -0,0 +1,44 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <ostream> +#include <assert.h> + + +namespace crimson { + namespace simple_scheduler { + + // since we send no additional data out + // NOTE: Change name to RespParams? Is it used elsewhere? + struct NullData { + friend std::ostream& operator<<(std::ostream& out, const NullData& n) { + out << "NullData{ EMPTY }"; + return out; + } + }; // struct NullData + + + struct ReqParams { + friend std::ostream& operator<<(std::ostream& out, const ReqParams& rp) { + out << "ReqParams{ EMPTY }"; + return out; + } + }; + + } +} diff --git a/src/dmclock/sim/src/ssched/ssched_server.h b/src/dmclock/sim/src/ssched/ssched_server.h new file mode 100644 index 000000000..c4e057a88 --- /dev/null +++ b/src/dmclock/sim/src/ssched/ssched_server.h @@ -0,0 +1,194 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + +#include <memory> +#include <mutex> +#include <deque> +#include <functional> + +#include "boost/variant.hpp" + +#include "ssched_recs.h" + +#ifdef PROFILE +#include "profile.h" +#endif + +namespace crimson { + + namespace simple_scheduler { + + template<typename C, typename R, typename Time> + class SimpleQueue { + + public: + + using RequestRef = std::unique_ptr<R>; + + // a function to see whether the server can handle another request + using CanHandleRequestFunc = std::function<bool(void)>; + + // a function to submit a request to the server; the second + // parameter is a callback when it's completed + using HandleRequestFunc = + std::function<void(const C&,RequestRef,NullData,uint64_t)>; + + struct PullReq { + enum class Type { returning, none }; + + struct Retn { + C client; + RequestRef request; + }; + + Type type; + boost::variant<Retn> data; + }; + + protected: + + enum class Mechanism { push, pull }; + + struct QRequest { + C client; + RequestRef request; + }; + + bool finishing = false; + Mechanism mechanism; + + CanHandleRequestFunc can_handle_f; + HandleRequestFunc handle_f; + + mutable std::mutex queue_mtx; + using DataGuard = std::lock_guard<decltype(queue_mtx)>; + + std::deque<QRequest> queue; + +#ifdef PROFILE + public: + ProfileTimer<std::chrono::nanoseconds> pull_request_timer; + ProfileTimer<std::chrono::nanoseconds> add_request_timer; + ProfileTimer<std::chrono::nanoseconds> request_complete_timer; + protected: +#endif + + public: + + // push full constructor + SimpleQueue(CanHandleRequestFunc _can_handle_f, + HandleRequestFunc _handle_f) : + mechanism(Mechanism::push), + can_handle_f(_can_handle_f), + handle_f(_handle_f) + { + // empty + } + + SimpleQueue() : + mechanism(Mechanism::pull) + { + // empty + } + + ~SimpleQueue() { + finishing = true; + } + + void add_request(R&& request, + const C& client_id, + const ReqParams& req_params, + uint64_t request_cost) { + add_request(RequestRef(new R(std::move(request))), + client_id, req_params, request_cost); + } + + void add_request(RequestRef&& request, + const C& client_id, + const ReqParams& req_params, + uint64_t request_cost) { + DataGuard g(queue_mtx); + +#ifdef PROFILE + add_request_timer.start(); +#endif + queue.emplace_back(QRequest{client_id, std::move(request)}); + + if (Mechanism::push == mechanism) { + schedule_request(); + } + +#ifdef PROFILE + add_request_timer.stop(); +#endif + } // add_request + + void request_completed() { + assert(Mechanism::push == mechanism); + DataGuard g(queue_mtx); + +#ifdef PROFILE + request_complete_timer.start(); +#endif + schedule_request(); + +#ifdef PROFILE + request_complete_timer.stop(); +#endif + } // request_completed + + PullReq pull_request() { + assert(Mechanism::pull == mechanism); + PullReq result; + DataGuard g(queue_mtx); + +#ifdef PROFILE + pull_request_timer.start(); +#endif + + if (queue.empty()) { + result.type = PullReq::Type::none; + } else { + auto front = queue.front(); + result.type = PullReq::Type::returning; + result.data = + typename PullReq::Retn{front.client, std::move(front.request)}; + queue.pop(); + } + +#ifdef PROFILE + pull_request_timer.stop(); +#endif + + return result; + } + + protected: + + // queue_mtx should be held when called; should only be called + // when mechanism is push + void schedule_request() { + if (!queue.empty() && can_handle_f()) { + auto& front = queue.front(); + static NullData null_data; + handle_f(front.client, std::move(front.request), null_data, 1u); + queue.pop_front(); + } + } + }; + }; +}; diff --git a/src/dmclock/sim/src/str_list.cc b/src/dmclock/sim/src/str_list.cc new file mode 100644 index 000000000..22109e008 --- /dev/null +++ b/src/dmclock/sim/src/str_list.cc @@ -0,0 +1,106 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab +/* + * Ceph - scalable distributed file system + * + * Copyright (C) 2009-2010 Dreamhost + * + * This is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License version 2.1, as published by the Free Software + * Foundation. See file COPYING. + * + */ + +#include "str_list.h" + +using std::string; +using std::vector; +using std::set; +using std::list; + +static bool get_next_token(const string &s, size_t& pos, const char *delims, string& token) +{ + int start = s.find_first_not_of(delims, pos); + int end; + + if (start < 0){ + pos = s.size(); + return false; + } + + end = s.find_first_of(delims, start); + if (end >= 0) + pos = end + 1; + else { + pos = end = s.size(); + } + + token = s.substr(start, end - start); + return true; +} + +void get_str_list(const string& str, const char *delims, list<string>& str_list) +{ + size_t pos = 0; + string token; + + str_list.clear(); + + while (pos < str.size()) { + if (get_next_token(str, pos, delims, token)) { + if (token.size() > 0) { + str_list.push_back(token); + } + } + } +} + +void get_str_list(const string& str, list<string>& str_list) +{ + const char *delims = ";,= \t"; + return get_str_list(str, delims, str_list); +} + +void get_str_vec(const string& str, const char *delims, vector<string>& str_vec) +{ + size_t pos = 0; + string token; + str_vec.clear(); + + while (pos < str.size()) { + if (get_next_token(str, pos, delims, token)) { + if (token.size() > 0) { + str_vec.push_back(token); + } + } + } +} + +void get_str_vec(const string& str, vector<string>& str_vec) +{ + const char *delims = ";,= \t"; + return get_str_vec(str, delims, str_vec); +} + +void get_str_set(const string& str, const char *delims, set<string>& str_set) +{ + size_t pos = 0; + string token; + + str_set.clear(); + + while (pos < str.size()) { + if (get_next_token(str, pos, delims, token)) { + if (token.size() > 0) { + str_set.insert(token); + } + } + } +} + +void get_str_set(const string& str, set<string>& str_set) +{ + const char *delims = ";,= \t"; + return get_str_set(str, delims, str_set); +} diff --git a/src/dmclock/sim/src/str_list.h b/src/dmclock/sim/src/str_list.h new file mode 100644 index 000000000..4a6dcc57f --- /dev/null +++ b/src/dmclock/sim/src/str_list.h @@ -0,0 +1,109 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2009 Red Hat Inc. + * + * Forked from Red Hat's Ceph project. + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#ifndef CEPH_STRLIST_H +#define CEPH_STRLIST_H + +#include <list> +#include <set> +#include <sstream> +#include <string> +#include <vector> + +/** + * Split **str** into a list of strings, using the ";,= \t" delimiters and output the result in **str_list**. + * + * @param [in] str String to split and save as list + * @param [out] str_list List modified containing str after it has been split +**/ +extern void get_str_list(const std::string& str, + std::list<std::string>& str_list); + +/** + * Split **str** into a list of strings, using the **delims** delimiters and output the result in **str_list**. + * + * @param [in] str String to split and save as list + * @param [in] delims characters used to split **str** + * @param [out] str_list List modified containing str after it has been split +**/ +extern void get_str_list(const std::string& str, + const char *delims, + std::list<std::string>& str_list); + +/** + * Split **str** into a list of strings, using the ";,= \t" delimiters and output the result in **str_vec**. + * + * @param [in] str String to split and save as Vector + * @param [out] str_vec Vector modified containing str after it has been split +**/ +extern void get_str_vec(const std::string& str, + std::vector<std::string>& str_vec); + +/** + * Split **str** into a list of strings, using the **delims** delimiters and output the result in **str_vec**. + * + * @param [in] str String to split and save as Vector + * @param [in] delims characters used to split **str** + * @param [out] str_vec Vector modified containing str after it has been split +**/ +extern void get_str_vec(const std::string& str, + const char *delims, + std::vector<std::string>& str_vec); + +/** + * Split **str** into a list of strings, using the ";,= \t" delimiters and output the result in **str_list**. + * + * @param [in] str String to split and save as Set + * @param [out] str_list Set modified containing str after it has been split +**/ +extern void get_str_set(const std::string& str, + std::set<std::string>& str_list); + +/** + * Split **str** into a list of strings, using the **delims** delimiters and output the result in **str_list**. + * + * @param [in] str String to split and save as Set + * @param [in] delims characters used to split **str** + * @param [out] str_list Set modified containing str after it has been split +**/ +extern void get_str_set(const std::string& str, + const char *delims, + std::set<std::string>& str_list); + +/** + * Return a String containing the vector **v** joined with **sep** + * + * If **v** is empty, the function returns an empty string + * For each element in **v**, + * it will concatenate this element and **sep** with result + * + * @param [in] v Vector to join as a String + * @param [in] sep String used to join each element from **v** + * @return empty string if **v** is empty or concatenated string +**/ +inline std::string str_join(const std::vector<std::string>& v, const std::string& sep) +{ + if (v.empty()) + return std::string(); + std::vector<std::string>::const_iterator i = v.begin(); + std::string r = *i; + for (++i; i != v.end(); ++i) { + r += sep; + r += *i; + } + return r; +} + +#endif diff --git a/src/dmclock/sim/src/test_dmclock.cc b/src/dmclock/sim/src/test_dmclock.cc new file mode 100644 index 000000000..402a47335 --- /dev/null +++ b/src/dmclock/sim/src/test_dmclock.cc @@ -0,0 +1,55 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + +#include <type_traits> + +#include "dmclock_recs.h" +#include "dmclock_server.h" +#include "dmclock_client.h" + +#include "sim_recs.h" +#include "sim_server.h" +#include "sim_client.h" + +#include "test_dmclock.h" + + +namespace test = crimson::test_dmc; + + +// Note: if this static_assert fails then our two definitions of Cost +// do not match; change crimson::qos_simulation::Cost to match the +// definition of crimson::dmclock::Cost. +static_assert(std::is_same<crimson::qos_simulation::Cost,crimson::dmclock::Cost>::value, + "Please make sure the simulator type crimson::qos_simulation::Cost matches the dmclock type crimson::dmclock::Cost."); + + +void test::dmc_server_accumulate_f(test::DmcAccum& a, + const test::dmc::PhaseType& phase) { + if (test::dmc::PhaseType::reservation == phase) { + ++a.reservation_count; + } else { + ++a.proportion_count; + } +} + + +void test::dmc_client_accumulate_f(test::DmcAccum& a, + const test::dmc::PhaseType& phase) { + if (test::dmc::PhaseType::reservation == phase) { + ++a.reservation_count; + } else { + ++a.proportion_count; + } +} diff --git a/src/dmclock/sim/src/test_dmclock.h b/src/dmclock/sim/src/test_dmclock.h new file mode 100644 index 000000000..1c7a55968 --- /dev/null +++ b/src/dmclock/sim/src/test_dmclock.h @@ -0,0 +1,64 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include "dmclock_recs.h" +#include "dmclock_server.h" +#include "dmclock_client.h" + +#include "sim_recs.h" +#include "sim_server.h" +#include "sim_client.h" + +#include "simulate.h" + + +namespace crimson { + namespace test_dmc { + + namespace dmc = crimson::dmclock; + namespace sim = crimson::qos_simulation; + + struct DmcAccum { + uint64_t reservation_count = 0; + uint64_t proportion_count = 0; + }; + + using DmcQueue = dmc::PushPriorityQueue<ClientId,sim::TestRequest>; + using DmcServiceTracker = dmc::ServiceTracker<ServerId,dmc::OrigTracker>; + + using DmcServer = sim::SimulatedServer<DmcQueue, + dmc::ReqParams, + dmc::PhaseType, + DmcAccum>; + + using DmcClient = sim::SimulatedClient<DmcServiceTracker, + dmc::ReqParams, + dmc::PhaseType, + DmcAccum>; + + using CreateQueueF = std::function<DmcQueue*(DmcQueue::CanHandleRequestFunc, + DmcQueue::HandleRequestFunc)>; + + using MySim = sim::Simulation<ServerId,ClientId,DmcServer,DmcClient>; + + using SubmitFunc = DmcClient::SubmitFunc; + + extern void dmc_server_accumulate_f(DmcAccum& a, + const dmc::PhaseType& phase); + + extern void dmc_client_accumulate_f(DmcAccum& a, + const dmc::PhaseType& phase); + } // namespace test_dmc +} // namespace crimson diff --git a/src/dmclock/sim/src/test_dmclock_main.cc b/src/dmclock/sim/src/test_dmclock_main.cc new file mode 100644 index 000000000..5cf656608 --- /dev/null +++ b/src/dmclock/sim/src/test_dmclock_main.cc @@ -0,0 +1,342 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include "test_dmclock.h" +#include "config.h" + +#ifdef PROFILE +#include "profile.h" +#endif + + +namespace dmc = crimson::dmclock; +namespace test = crimson::test_dmc; +namespace sim = crimson::qos_simulation; + +using namespace std::placeholders; + + +namespace crimson { + namespace test_dmc { + void server_data(std::ostream& out, + test::MySim* sim, + test::MySim::ServerFilter server_disp_filter, + int head_w, int data_w, int data_prec); + + void client_data(std::ostream& out, + test::MySim* sim, + test::MySim::ClientFilter client_disp_filter, + int head_w, int data_w, int data_prec); + } +} + + +int main(int argc, char* argv[]) { + std::vector<const char*> args; + for (int i = 1; i < argc; ++i) { + args.push_back(argv[i]); + } + + std::string conf_file_list; + sim::ceph_argparse_early_args(args, &conf_file_list); + + sim::sim_config_t g_conf; + std::vector<sim::cli_group_t> &cli_group = g_conf.cli_group; + std::vector<sim::srv_group_t> &srv_group = g_conf.srv_group; + + if (!conf_file_list.empty()) { + int ret; + ret = sim::parse_config_file(conf_file_list, g_conf); + if (ret) { + // error + _exit(1); + } + } else { + // default simulation parameter + g_conf.client_groups = 2; + + sim::srv_group_t st; + srv_group.push_back(st); + + sim::cli_group_t ct1(99, 0); + cli_group.push_back(ct1); + + sim::cli_group_t ct2(1, 10); + cli_group.push_back(ct2); + } + + const unsigned server_groups = g_conf.server_groups; + const unsigned client_groups = g_conf.client_groups; + const bool server_random_selection = g_conf.server_random_selection; + const bool server_soft_limit = g_conf.server_soft_limit; + const double anticipation_timeout = g_conf.anticipation_timeout; + unsigned server_total_count = 0; + unsigned client_total_count = 0; + + for (unsigned i = 0; i < client_groups; ++i) { + client_total_count += cli_group[i].client_count; + } + + for (unsigned i = 0; i < server_groups; ++i) { + server_total_count += srv_group[i].server_count; + } + + std::vector<test::dmc::ClientInfo> client_info; + for (unsigned i = 0; i < client_groups; ++i) { + client_info.push_back(test::dmc::ClientInfo + { cli_group[i].client_reservation, + cli_group[i].client_weight, + cli_group[i].client_limit } ); + } + + auto ret_client_group_f = [&](const ClientId& c) -> unsigned { + unsigned group_max = 0; + unsigned i = 0; + for (; i < client_groups; ++i) { + group_max += cli_group[i].client_count; + if (c < group_max) { + break; + } + } + return i; + }; + + auto ret_server_group_f = [&](const ServerId& s) -> unsigned { + unsigned group_max = 0; + unsigned i = 0; + for (; i < server_groups; ++i) { + group_max += srv_group[i].server_count; + if (s < group_max) { + break; + } + } + return i; + }; + + auto client_info_f = + [=](const ClientId& c) -> const test::dmc::ClientInfo* { + return &client_info[ret_client_group_f(c)]; + }; + + auto client_disp_filter = [=] (const ClientId& i) -> bool { + return i < 3 || i >= (client_total_count - 3); + }; + + auto server_disp_filter = [=] (const ServerId& i) -> bool { + return i < 3 || i >= (server_total_count - 3); + }; + + + test::MySim *simulation; + + + // lambda to post a request to the identified server; called by client + test::SubmitFunc server_post_f = + [&simulation, + &cli_group, + &ret_client_group_f](const ServerId& server, + sim::TestRequest&& request, + const ClientId& client_id, + const test::dmc::ReqParams& req_params) { + test::DmcServer& s = simulation->get_server(server); + sim::Cost request_cost = cli_group[ret_client_group_f(client_id)].client_req_cost; + s.post(std::move(request), client_id, req_params, request_cost); + }; + + std::vector<std::vector<sim::CliInst>> cli_inst; + for (unsigned i = 0; i < client_groups; ++i) { + if (cli_group[i].client_wait == std::chrono::seconds(0)) { + cli_inst.push_back( + { { sim::req_op, + (uint32_t)cli_group[i].client_total_ops, + (double)cli_group[i].client_iops_goal, + (uint16_t)cli_group[i].client_outstanding_ops } } ); + } else { + cli_inst.push_back( + { { sim::wait_op, cli_group[i].client_wait }, + { sim::req_op, + (uint32_t)cli_group[i].client_total_ops, + (double)cli_group[i].client_iops_goal, + (uint16_t)cli_group[i].client_outstanding_ops } } ); + } + } + + simulation = new test::MySim(); + + test::DmcServer::ClientRespFunc client_response_f = + [&simulation](ClientId client_id, + const sim::TestResponse& resp, + const ServerId& server_id, + const dmc::PhaseType& phase, + const sim::Cost request_cost) { + simulation->get_client(client_id).receive_response(resp, + server_id, + phase, + request_cost); + }; + + test::CreateQueueF create_queue_f = + [&](test::DmcQueue::CanHandleRequestFunc can_f, + test::DmcQueue::HandleRequestFunc handle_f) -> test::DmcQueue* { + return new test::DmcQueue(client_info_f, + can_f, + handle_f, + server_soft_limit ? dmc::AtLimit::Allow : dmc::AtLimit::Wait, + anticipation_timeout); + }; + + + auto create_server_f = [&](ServerId id) -> test::DmcServer* { + unsigned i = ret_server_group_f(id); + return new test::DmcServer(id, + srv_group[i].server_iops, + srv_group[i].server_threads, + client_response_f, + test::dmc_server_accumulate_f, + create_queue_f); + }; + + auto create_client_f = [&](ClientId id) -> test::DmcClient* { + unsigned i = ret_client_group_f(id); + test::MySim::ClientBasedServerSelectFunc server_select_f; + unsigned client_server_select_range = cli_group[i].client_server_select_range; + if (!server_random_selection) { + server_select_f = simulation->make_server_select_alt_range(client_server_select_range); + } else { + server_select_f = simulation->make_server_select_ran_range(client_server_select_range); + } + return new test::DmcClient(id, + server_post_f, + std::bind(server_select_f, _1, id), + test::dmc_client_accumulate_f, + cli_inst[i]); + }; + +#if 1 + std::cout << "[global]" << std::endl << g_conf << std::endl; + for (unsigned i = 0; i < client_groups; ++i) { + std::cout << std::endl << "[client." << i << "]" << std::endl; + std::cout << cli_group[i] << std::endl; + } + for (unsigned i = 0; i < server_groups; ++i) { + std::cout << std::endl << "[server." << i << "]" << std::endl; + std::cout << srv_group[i] << std::endl; + } + std::cout << std::endl; +#endif + + simulation->add_servers(server_total_count, create_server_f); + simulation->add_clients(client_total_count, create_client_f); + + simulation->run(); + simulation->display_stats(std::cout, + &test::server_data, &test::client_data, + server_disp_filter, client_disp_filter); + + delete simulation; +} // main + + +void test::client_data(std::ostream& out, + test::MySim* sim, + test::MySim::ClientFilter client_disp_filter, + int head_w, int data_w, int data_prec) { + // report how many ops were done by reservation and proportion for + // each client + + int total_r = 0; + out << std::setw(head_w) << "res_ops:"; + for (unsigned i = 0; i < sim->get_client_count(); ++i) { + const auto& client = sim->get_client(i); + auto r = client.get_accumulator().reservation_count; + total_r += r; + if (!client_disp_filter(i)) continue; + out << " " << std::setw(data_w) << r; + } + out << " " << std::setw(data_w) << std::setprecision(data_prec) << + std::fixed << total_r << std::endl; + + int total_p = 0; + out << std::setw(head_w) << "prop_ops:"; + for (unsigned i = 0; i < sim->get_client_count(); ++i) { + const auto& client = sim->get_client(i); + auto p = client.get_accumulator().proportion_count; + total_p += p; + if (!client_disp_filter(i)) continue; + out << " " << std::setw(data_w) << p; + } + out << " " << std::setw(data_w) << std::setprecision(data_prec) << + std::fixed << total_p << std::endl; +} + + +void test::server_data(std::ostream& out, + test::MySim* sim, + test::MySim::ServerFilter server_disp_filter, + int head_w, int data_w, int data_prec) { + out << std::setw(head_w) << "res_ops:"; + int total_r = 0; + for (unsigned i = 0; i < sim->get_server_count(); ++i) { + const auto& server = sim->get_server(i); + auto rc = server.get_accumulator().reservation_count; + total_r += rc; + if (!server_disp_filter(i)) continue; + out << " " << std::setw(data_w) << rc; + } + out << " " << std::setw(data_w) << std::setprecision(data_prec) << + std::fixed << total_r << std::endl; + + out << std::setw(head_w) << "prop_ops:"; + int total_p = 0; + for (unsigned i = 0; i < sim->get_server_count(); ++i) { + const auto& server = sim->get_server(i); + auto pc = server.get_accumulator().proportion_count; + total_p += pc; + if (!server_disp_filter(i)) continue; + out << " " << std::setw(data_w) << pc; + } + out << " " << std::setw(data_w) << std::setprecision(data_prec) << + std::fixed << total_p << std::endl; + + const auto& q = sim->get_server(0).get_priority_queue(); + out << std::endl << + " k-way heap: " << q.get_heap_branching_factor() << std::endl + << std::endl; + +#ifdef PROFILE + crimson::ProfileCombiner<std::chrono::nanoseconds> art_combiner; + crimson::ProfileCombiner<std::chrono::nanoseconds> rct_combiner; + for (unsigned i = 0; i < sim->get_server_count(); ++i) { + const auto& q = sim->get_server(i).get_priority_queue(); + const auto& art = q.add_request_timer; + art_combiner.combine(art); + const auto& rct = q.request_complete_timer; + rct_combiner.combine(rct); + } + out << "Server add_request_timer: count:" << art_combiner.get_count() << + ", mean:" << art_combiner.get_mean() << + ", std_dev:" << art_combiner.get_std_dev() << + ", low:" << art_combiner.get_low() << + ", high:" << art_combiner.get_high() << std::endl; + out << "Server request_complete_timer: count:" << rct_combiner.get_count() << + ", mean:" << rct_combiner.get_mean() << + ", std_dev:" << rct_combiner.get_std_dev() << + ", low:" << rct_combiner.get_low() << + ", high:" << rct_combiner.get_high() << std::endl; + out << "Server combined mean: " << + (art_combiner.get_mean() + rct_combiner.get_mean()) << + std::endl; +#endif +} diff --git a/src/dmclock/sim/src/test_ssched.cc b/src/dmclock/sim/src/test_ssched.cc new file mode 100644 index 000000000..b06273dc0 --- /dev/null +++ b/src/dmclock/sim/src/test_ssched.cc @@ -0,0 +1,40 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include "ssched_recs.h" +#include "ssched_server.h" +#include "ssched_client.h" + +#include "sim_recs.h" +#include "sim_server.h" +#include "sim_client.h" + +#include "test_ssched.h" + + +namespace test = crimson::test_simple_scheduler; +namespace ssched = crimson::simple_scheduler; + + +void test::simple_server_accumulate_f(test::SimpleAccum& a, + const ssched::NullData& add_info) { + ++a.request_count; +} + + +void test::simple_client_accumulate_f(test::SimpleAccum& a, + const ssched::NullData& ignore) { + // empty +} diff --git a/src/dmclock/sim/src/test_ssched.h b/src/dmclock/sim/src/test_ssched.h new file mode 100644 index 000000000..0d778709a --- /dev/null +++ b/src/dmclock/sim/src/test_ssched.h @@ -0,0 +1,64 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include "ssched_server.h" +#include "ssched_client.h" + +#include "sim_recs.h" +#include "sim_server.h" +#include "sim_client.h" + +#include "simulate.h" + + +namespace crimson { + namespace test_simple_scheduler { + + namespace ssched = crimson::simple_scheduler; + namespace sim = crimson::qos_simulation; + + using Time = double; + + struct SimpleAccum { + uint32_t request_count = 0; + }; + + using SimpleQueue = ssched::SimpleQueue<ClientId,sim::TestRequest,Time>; + + using SimpleServer = sim::SimulatedServer<SimpleQueue, + ssched::ReqParams, + ssched::NullData, + SimpleAccum>; + using SimpleClient = sim::SimulatedClient<ssched::ServiceTracker<ServerId>, + ssched::ReqParams, + ssched::NullData, + SimpleAccum>; + + using CreateQueueF = + std::function<SimpleQueue*(SimpleQueue::CanHandleRequestFunc, + SimpleQueue::HandleRequestFunc)>; + + + using MySim = sim::Simulation<ServerId,ClientId,SimpleServer,SimpleClient>; + + using SubmitFunc = SimpleClient::SubmitFunc; + + extern void simple_server_accumulate_f(SimpleAccum& a, + const ssched::NullData& add_info); + + extern void simple_client_accumulate_f(SimpleAccum& a, + const ssched::NullData& ignore); + } // namespace test_simple +} // namespace crimson diff --git a/src/dmclock/sim/src/test_ssched_main.cc b/src/dmclock/sim/src/test_ssched_main.cc new file mode 100644 index 000000000..ace4f8cce --- /dev/null +++ b/src/dmclock/sim/src/test_ssched_main.cc @@ -0,0 +1,199 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include "test_ssched.h" + + +#ifdef PROFILE +#include "profile.h" +#endif + + +namespace test = crimson::test_simple_scheduler; +namespace ssched = crimson::simple_scheduler; +namespace sim = crimson::qos_simulation; + +using namespace std::placeholders; + + +namespace crimson { + namespace test_simple_scheduler { + void client_data(std::ostream& out, + test::MySim* sim, + test::MySim::ClientFilter client_disp_filter, + int head_w, int data_w, int data_prec); + + void server_data(std::ostream& out, + test::MySim* sim, + test::MySim::ServerFilter server_disp_filter, + int head_w, int data_w, int data_prec); + } // namespace test_simple +} // namespace crimson + + +using Cost = uint32_t; + + +int main(int argc, char* argv[]) { + // server params + + const unsigned server_count = 100; + const unsigned server_iops = 40; + const unsigned server_threads = 1; + + // client params + + const unsigned client_total_ops = 1000; + const unsigned client_count = 100; + const unsigned client_server_select_range = 10; + const unsigned client_wait_count = 1; + const unsigned client_iops_goal = 50; + const unsigned client_outstanding_ops = 100; + const std::chrono::seconds client_wait(10); + + auto client_disp_filter = [=] (const ClientId& i) -> bool { + return i < 3 || i >= (client_count - 3); + }; + + auto server_disp_filter = [=] (const ServerId& i) -> bool { + return i < 3 || i >= (server_count - 3); + }; + + + test::MySim *simulation; + + // lambda to post a request to the identified server; called by client + test::SubmitFunc server_post_f = + [&simulation](const ServerId& server_id, + sim::TestRequest&& request, + const ClientId& client_id, + const ssched::ReqParams& req_params) { + auto& server = simulation->get_server(server_id); + server.post(std::move(request), client_id, req_params, 1u); + }; + + static std::vector<sim::CliInst> no_wait = + { { sim::req_op, client_total_ops, client_iops_goal, client_outstanding_ops } }; + static std::vector<sim::CliInst> wait = + { { sim::wait_op, client_wait }, + { sim::req_op, client_total_ops, client_iops_goal, client_outstanding_ops } }; + + simulation = new test::MySim(); + +#if 1 + test::MySim::ClientBasedServerSelectFunc server_select_f = + simulation->make_server_select_alt_range(client_server_select_range); +#elif 0 + test::MySim::ClientBasedServerSelectFunc server_select_f = + std::bind(&test::MySim::server_select_random, simulation, _1, _2); +#else + test::MySim::ClientBasedServerSelectFunc server_select_f = + std::bind(&test::MySim::server_select_0, simulation, _1, _2); +#endif + + test::SimpleServer::ClientRespFunc client_response_f = + [&simulation](ClientId client_id, + const sim::TestResponse& resp, + const ServerId& server_id, + const ssched::NullData& resp_params, + Cost request_cost) { + simulation->get_client(client_id).receive_response(resp, + server_id, + resp_params, + request_cost); + }; + + test::CreateQueueF create_queue_f = + [&](test::SimpleQueue::CanHandleRequestFunc can_f, + test::SimpleQueue::HandleRequestFunc handle_f) -> test::SimpleQueue* { + return new test::SimpleQueue(can_f, handle_f); + }; + + auto create_server_f = [&](ServerId id) -> test::SimpleServer* { + return new test::SimpleServer(id, + server_iops, server_threads, + client_response_f, + test::simple_server_accumulate_f, + create_queue_f); + }; + + auto create_client_f = [&](ClientId id) -> test::SimpleClient* { + return new test::SimpleClient(id, + server_post_f, + std::bind(server_select_f, _1, id), + test::simple_client_accumulate_f, + id < (client_count - client_wait_count) + ? no_wait : wait); + }; + + simulation->add_servers(server_count, create_server_f); + simulation->add_clients(client_count, create_client_f); + + simulation->run(); + simulation->display_stats(std::cout, + &test::server_data, &test::client_data, + server_disp_filter, client_disp_filter); +} // main + + +void test::client_data(std::ostream& out, + test::MySim* sim, + test::MySim::ClientFilter client_disp_filter, + int head_w, int data_w, int data_prec) { + // empty +} + + +void test::server_data(std::ostream& out, + test::MySim* sim, + test::MySim::ServerFilter server_disp_filter, + int head_w, int data_w, int data_prec) { + out << std::setw(head_w) << "requests:"; + int total_req = 0; + for (unsigned i = 0; i < sim->get_server_count(); ++i) { + const auto& server = sim->get_server(i); + auto req_count = server.get_accumulator().request_count; + total_req += req_count; + if (!server_disp_filter(i)) continue; + out << std::setw(data_w) << req_count; + } + out << std::setw(data_w) << std::setprecision(data_prec) << + std::fixed << total_req << std::endl; + +#ifdef PROFILE + crimson::ProfileCombiner<std::chrono::nanoseconds> art_combiner; + crimson::ProfileCombiner<std::chrono::nanoseconds> rct_combiner; + for (unsigned i = 0; i < sim->get_server_count(); ++i) { + const auto& q = sim->get_server(i).get_priority_queue(); + const auto& art = q.add_request_timer; + art_combiner.combine(art); + const auto& rct = q.request_complete_timer; + rct_combiner.combine(rct); + } + out << "Server add_request_timer: count:" << art_combiner.get_count() << + ", mean:" << art_combiner.get_mean() << + ", std_dev:" << art_combiner.get_std_dev() << + ", low:" << art_combiner.get_low() << + ", high:" << art_combiner.get_high() << std::endl; + out << "Server request_complete_timer: count:" << rct_combiner.get_count() << + ", mean:" << rct_combiner.get_mean() << + ", std_dev:" << rct_combiner.get_std_dev() << + ", low:" << rct_combiner.get_low() << + ", high:" << rct_combiner.get_high() << std::endl; + out << "Server combined mean: " << + (art_combiner.get_mean() + rct_combiner.get_mean()) << + std::endl; +#endif +} diff --git a/src/dmclock/src/CMakeLists.txt b/src/dmclock/src/CMakeLists.txt new file mode 100644 index 000000000..d13229e40 --- /dev/null +++ b/src/dmclock/src/CMakeLists.txt @@ -0,0 +1,14 @@ +set(dmc_srcs dmclock_util.cc ../support/src/run_every.cc) + +add_library(dmclock STATIC ${dmc_srcs}) +add_library(dmclock::dmclock ALIAS dmclock) + +target_compile_options(dmclock PRIVATE + "-Wno-write-strings" "-Wall") +target_include_directories(dmclock PUBLIC + $<BUILD_INTERFACE:${PROJECT_SOURCE_DIR}/src> + $<BUILD_INTERFACE:${PROJECT_SOURCE_DIR}/support/src>) + +target_link_libraries(dmclock + PUBLIC Boost::boost + PRIVATE Threads::Threads) diff --git a/src/dmclock/src/dmclock_client.h b/src/dmclock/src/dmclock_client.h new file mode 100644 index 000000000..1143147ce --- /dev/null +++ b/src/dmclock/src/dmclock_client.h @@ -0,0 +1,289 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2017 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + +#include <map> +#include <deque> +#include <chrono> +#include <thread> +#include <mutex> +#include <condition_variable> + +#include "../support/src/run_every.h" +#include "dmclock_util.h" +#include "dmclock_recs.h" + + +namespace crimson { + namespace dmclock { + + // OrigTracker is a best-effort implementation of the the original + // dmClock calculations of delta and rho. It adheres to an + // interface, implemented via a template type, that allows it to + // be replaced with an alternative. The interface consists of the + // static create, prepare_req, resp_update, and get_last_delta + // functions. + class OrigTracker { + Counter delta_prev_req; + Counter rho_prev_req; + uint32_t my_delta; + uint32_t my_rho; + + public: + + OrigTracker(Counter global_delta, + Counter global_rho) : + delta_prev_req(global_delta), + rho_prev_req(global_rho), + my_delta(0), + my_rho(0) + { /* empty */ } + + static inline OrigTracker create(Counter the_delta, Counter the_rho) { + return OrigTracker(the_delta, the_rho); + } + + inline ReqParams prepare_req(Counter& the_delta, Counter& the_rho) { + Counter delta_out = the_delta - delta_prev_req - my_delta; + Counter rho_out = the_rho - rho_prev_req - my_rho; + delta_prev_req = the_delta; + rho_prev_req = the_rho; + my_delta = 0; + my_rho = 0; + return ReqParams(uint32_t(delta_out), uint32_t(rho_out)); + } + + inline void resp_update(PhaseType phase, + Counter& the_delta, + Counter& the_rho, + Cost cost) { + the_delta += cost; + my_delta += cost; + if (phase == PhaseType::reservation) { + the_rho += cost; + my_rho += cost; + } + } + + inline Counter get_last_delta() const { + return delta_prev_req; + } + }; // struct OrigTracker + + + // BorrowingTracker always returns a positive delta and rho. If + // not enough responses have come in to allow that, we will borrow + // a future response and repay it later. + class BorrowingTracker { + Counter delta_prev_req; + Counter rho_prev_req; + Counter delta_borrow; + Counter rho_borrow; + + public: + + BorrowingTracker(Counter global_delta, Counter global_rho) : + delta_prev_req(global_delta), + rho_prev_req(global_rho), + delta_borrow(0), + rho_borrow(0) + { /* empty */ } + + static inline BorrowingTracker create(Counter the_delta, + Counter the_rho) { + return BorrowingTracker(the_delta, the_rho); + } + + inline Counter calc_with_borrow(const Counter& global, + const Counter& previous, + Counter& borrow) { + Counter result = global - previous; + if (0 == result) { + // if no replies have come in, borrow one from the future + ++borrow; + return 1; + } else if (result > borrow) { + // if we can give back all of what we borrowed, do so + result -= borrow; + borrow = 0; + return result; + } else { + // can only return part of what was borrowed in order to + // return positive + borrow = borrow - result + 1; + return 1; + } + } + + inline ReqParams prepare_req(Counter& the_delta, Counter& the_rho) { + Counter delta_out = + calc_with_borrow(the_delta, delta_prev_req, delta_borrow); + Counter rho_out = + calc_with_borrow(the_rho, rho_prev_req, rho_borrow); + delta_prev_req = the_delta; + rho_prev_req = the_rho; + return ReqParams(uint32_t(delta_out), uint32_t(rho_out)); + } + + inline void resp_update(PhaseType phase, + Counter& the_delta, + Counter& the_rho, + Counter cost) { + the_delta += cost; + if (phase == PhaseType::reservation) { + the_rho += cost; + } + } + + inline Counter get_last_delta() const { + return delta_prev_req; + } + }; // struct BorrowingTracker + + + /* + * S is server identifier type + * + * T is the server info class that adheres to ServerTrackerIfc + * interface + */ + template<typename S, typename T = OrigTracker> + class ServiceTracker { + // we don't want to include gtest.h just for FRIEND_TEST + friend class dmclock_client_server_erase_Test; + + using TimePoint = decltype(std::chrono::steady_clock::now()); + using Duration = std::chrono::milliseconds; + using MarkPoint = std::pair<TimePoint,Counter>; + + Counter delta_counter; // # reqs completed + Counter rho_counter; // # reqs completed via reservation + std::map<S,T> server_map; + mutable std::mutex data_mtx; // protects Counters and map + + using DataGuard = std::lock_guard<decltype(data_mtx)>; + + // clean config + + std::deque<MarkPoint> clean_mark_points; + Duration clean_age; // age at which server tracker cleaned + + // NB: All threads declared at end, so they're destructed firs! + + std::unique_ptr<RunEvery> cleaning_job; + + + public: + + // we have to start the counters at 1, as 0 is used in the + // cleaning process + template<typename Rep, typename Per> + ServiceTracker(std::chrono::duration<Rep,Per> _clean_every, + std::chrono::duration<Rep,Per> _clean_age) : + delta_counter(1), + rho_counter(1), + clean_age(std::chrono::duration_cast<Duration>(_clean_age)) + { + cleaning_job = + std::unique_ptr<RunEvery>( + new RunEvery(_clean_every, + std::bind(&ServiceTracker::do_clean, this))); + } + + + // the reason we're overloading the constructor rather than + // using default values for the arguments is so that callers + // have to either use all defaults or specify all timings; with + // default arguments they could specify some without others + ServiceTracker() : + ServiceTracker(std::chrono::minutes(5), std::chrono::minutes(10)) + { + // empty + } + + + /* + * Incorporates the response data received into the counters. + */ + void track_resp(const S& server_id, + const PhaseType& phase, + Counter request_cost = 1u) { + DataGuard g(data_mtx); + + auto it = server_map.find(server_id); + if (server_map.end() == it) { + // this code can only run if a request did not precede the + // response or if the record was cleaned up b/w when + // the request was made and now + auto i = server_map.emplace(server_id, + T::create(delta_counter, rho_counter)); + it = i.first; + } + it->second.resp_update(phase, delta_counter, rho_counter, request_cost); + } + + /* + * Returns the ReqParams for the given server. + */ + ReqParams get_req_params(const S& server) { + DataGuard g(data_mtx); + auto it = server_map.find(server); + if (server_map.end() == it) { + server_map.emplace(server, + T::create(delta_counter, rho_counter)); + return ReqParams(1, 1); + } else { + return it->second.prepare_req(delta_counter, rho_counter); + } + } + + private: + + /* + * This is being called regularly by RunEvery. Every time it's + * called it notes the time and delta counter (mark point) in a + * deque. It also looks at the deque to find the most recent + * mark point that is older than clean_age. It then walks the + * map and delete all server entries that were last used before + * that mark point. + */ + void do_clean() { + TimePoint now = std::chrono::steady_clock::now(); + DataGuard g(data_mtx); + clean_mark_points.emplace_back(MarkPoint(now, delta_counter)); + + Counter earliest = 0; + auto point = clean_mark_points.front(); + while (point.first <= now - clean_age) { + earliest = point.second; + clean_mark_points.pop_front(); + point = clean_mark_points.front(); + } + + if (earliest > 0) { + for (auto i = server_map.begin(); + i != server_map.end(); + /* empty */) { + auto i2 = i++; + if (i2->second.get_last_delta() <= earliest) { + server_map.erase(i2); + } + } + } + } // do_clean + }; // class ServiceTracker + } +} diff --git a/src/dmclock/src/dmclock_recs.h b/src/dmclock/src/dmclock_recs.h new file mode 100644 index 000000000..a7dc44108 --- /dev/null +++ b/src/dmclock/src/dmclock_recs.h @@ -0,0 +1,74 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2017 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <ostream> +#include <assert.h> + + +namespace crimson { + namespace dmclock { + using Counter = uint64_t; + + // we're abstracting cost to its own type to better allow for + // future changes; we're assuming that Cost is relatively small + // and that it would be more efficient to pass-by-value than + // by-reference. + using Cost = uint32_t; + + enum class PhaseType : uint8_t { reservation, priority }; + + inline std::ostream& operator<<(std::ostream& out, const PhaseType& phase) { + out << (PhaseType::reservation == phase ? "reservation" : "priority"); + return out; + } + + struct ReqParams { + // count of all replies since last request + uint32_t delta; + + // count of reservation replies since last request + uint32_t rho; + + ReqParams(uint32_t _delta, uint32_t _rho) : + delta(_delta), + rho(_rho) + { + assert(rho <= delta); + } + + ReqParams() : + ReqParams(0, 0) + { + // empty + } + + ReqParams(const ReqParams& other) : + delta(other.delta), + rho(other.rho) + { + // empty + } + + friend std::ostream& operator<<(std::ostream& out, const ReqParams& rp) { + out << "ReqParams{ delta:" << rp.delta << + ", rho:" << rp.rho << " }"; + return out; + } + }; // class ReqParams + } +} diff --git a/src/dmclock/src/dmclock_server.h b/src/dmclock/src/dmclock_server.h new file mode 100644 index 000000000..79e7abcef --- /dev/null +++ b/src/dmclock/src/dmclock_server.h @@ -0,0 +1,1799 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2017 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + +/* COMPILATION OPTIONS + * + * The prop_heap does not seem to be necessary. The only thing it + * would help with is quickly finding the minimum proportion/prioity + * when an idle client became active. To have the code maintain the + * proportional heap, define USE_PROP_HEAP (i.e., compiler argument + * -DUSE_PROP_HEAP). + */ + +#include <assert.h> + +#include <cmath> +#include <memory> +#include <map> +#include <deque> +#include <queue> +#include <atomic> +#include <mutex> +#include <condition_variable> +#include <thread> +#include <iostream> +#include <sstream> +#include <limits> + +#include <boost/variant.hpp> + +#include "indirect_intrusive_heap.h" +#include "../support/src/run_every.h" +#include "dmclock_util.h" +#include "dmclock_recs.h" + +#ifdef PROFILE +#include "profile.h" +#endif + + +namespace crimson { + + namespace dmclock { + + namespace c = crimson; + + constexpr double max_tag = std::numeric_limits<double>::is_iec559 ? + std::numeric_limits<double>::infinity() : + std::numeric_limits<double>::max(); + constexpr double min_tag = std::numeric_limits<double>::is_iec559 ? + -std::numeric_limits<double>::infinity() : + std::numeric_limits<double>::lowest(); + constexpr unsigned tag_modulo = 1000000; + + constexpr auto standard_idle_age = std::chrono::seconds(300); + constexpr auto standard_erase_age = std::chrono::seconds(600); + constexpr auto standard_check_time = std::chrono::seconds(60); + constexpr auto aggressive_check_time = std::chrono::seconds(5); + constexpr unsigned standard_erase_max = 2000; + + enum class AtLimit { + // requests are delayed until the limit is restored + Wait, + // requests are allowed to exceed their limit, if all other reservations + // are met and below their limits + Allow, + // if an incoming request would exceed its limit, add_request() will + // reject it with EAGAIN instead of adding it to the queue. cannot be used + // with DelayedTagCalc, because add_request() needs an accurate tag + Reject, + }; + + // when AtLimit::Reject is used, only start rejecting requests once their + // limit is above this threshold. requests under this threshold are + // enqueued and processed like AtLimit::Wait + using RejectThreshold = Time; + + // the AtLimit constructor parameter can either accept AtLimit or a value + // for RejectThreshold (which implies AtLimit::Reject) + using AtLimitParam = boost::variant<AtLimit, RejectThreshold>; + + struct ClientInfo { + double reservation; // minimum + double weight; // proportional + double limit; // maximum + + // multiplicative inverses of above, which we use in calculations + // and don't want to recalculate repeatedly + double reservation_inv; + double weight_inv; + double limit_inv; + + // order parameters -- min, "normal", max + ClientInfo(double _reservation, double _weight, double _limit) { + update(_reservation, _weight, _limit); + } + + inline void update(double _reservation, double _weight, double _limit) { + reservation = _reservation; + weight = _weight; + limit = _limit; + reservation_inv = (0.0 == reservation) ? 0.0 : 1.0 / reservation; + weight_inv = (0.0 == weight) ? 0.0 : 1.0 / weight; + limit_inv = (0.0 == limit) ? 0.0 : 1.0 / limit; + } + + friend std::ostream& operator<<(std::ostream& out, + const ClientInfo& client) { + out << + "{ ClientInfo:: r:" << client.reservation << + " w:" << std::fixed << client.weight << + " l:" << std::fixed << client.limit << + " 1/r:" << std::fixed << client.reservation_inv << + " 1/w:" << std::fixed << client.weight_inv << + " 1/l:" << std::fixed << client.limit_inv << + " }"; + return out; + } + }; // class ClientInfo + + + struct RequestTag { + double reservation; + double proportion; + double limit; + uint32_t delta; + uint32_t rho; + Cost cost; + bool ready; // true when within limit + Time arrival; + + RequestTag(const RequestTag& prev_tag, + const ClientInfo& client, + const uint32_t _delta, + const uint32_t _rho, + const Time time, + const Cost _cost = 1u, + const double anticipation_timeout = 0.0) : + delta(_delta), + rho(_rho), + cost(_cost), + ready(false), + arrival(time) + { + assert(cost > 0); + Time max_time = time; + if (time - anticipation_timeout < prev_tag.arrival) + max_time -= anticipation_timeout; + + reservation = tag_calc(max_time, + prev_tag.reservation, + client.reservation_inv, + rho, + true, + cost); + proportion = tag_calc(max_time, + prev_tag.proportion, + client.weight_inv, + delta, + true, + cost); + limit = tag_calc(max_time, + prev_tag.limit, + client.limit_inv, + delta, + false, + cost); + + assert(reservation < max_tag || proportion < max_tag); + } + + RequestTag(const RequestTag& prev_tag, + const ClientInfo& client, + const ReqParams req_params, + const Time time, + const Cost cost = 1u, + const double anticipation_timeout = 0.0) : + RequestTag(prev_tag, client, req_params.delta, req_params.rho, time, + cost, anticipation_timeout) + { /* empty */ } + + RequestTag(const double _res, const double _prop, const double _lim, + const Time _arrival, + const uint32_t _delta = 0, + const uint32_t _rho = 0, + const Cost _cost = 1u) : + reservation(_res), + proportion(_prop), + limit(_lim), + delta(_delta), + rho(_rho), + cost(_cost), + ready(false), + arrival(_arrival) + { + assert(cost > 0); + assert(reservation < max_tag || proportion < max_tag); + } + + RequestTag(const RequestTag& other) : + reservation(other.reservation), + proportion(other.proportion), + limit(other.limit), + delta(other.delta), + rho(other.rho), + cost(other.cost), + ready(other.ready), + arrival(other.arrival) + { /* empty */ } + + static std::string format_tag_change(double before, double after) { + if (before == after) { + return std::string("same"); + } else { + std::stringstream ss; + ss << format_tag(before) << "=>" << format_tag(after); + return ss.str(); + } + } + + static std::string format_tag(double value) { + if (max_tag == value) { + return std::string("max"); + } else if (min_tag == value) { + return std::string("min"); + } else { + return format_time(value, tag_modulo); + } + } + + private: + + static double tag_calc(const Time time, + const double prev, + const double increment, + const uint32_t dist_req_val, + const bool extreme_is_high, + const Cost cost) { + if (0.0 == increment) { + return extreme_is_high ? max_tag : min_tag; + } else { + // insure 64-bit arithmetic before conversion to double + double tag_increment = increment * (uint64_t(dist_req_val) + cost); + return std::max(time, prev + tag_increment); + } + } + + friend std::ostream& operator<<(std::ostream& out, + const RequestTag& tag) { + out << + "{ RequestTag:: ready:" << (tag.ready ? "true" : "false") << + " r:" << format_tag(tag.reservation) << + " p:" << format_tag(tag.proportion) << + " l:" << format_tag(tag.limit) << +#if 0 // try to resolve this to make sure Time is operator<<'able. + " arrival:" << tag.arrival << +#endif + " }"; + return out; + } + }; // class RequestTag + + // C is client identifier type, R is request type, + // IsDelayed controls whether tag calculation is delayed until the request + // reaches the front of its queue. This is an optimization over the + // originally published dmclock algorithm, allowing it to use the most + // recent values of rho and delta. + // U1 determines whether to use client information function dynamically, + // B is heap branching factor + template<typename C, typename R, bool IsDelayed, bool U1, unsigned B> + class PriorityQueueBase { + // we don't want to include gtest.h just for FRIEND_TEST + friend class dmclock_server_client_idle_erase_Test; + + // types used for tag dispatch to select between implementations + using TagCalc = std::integral_constant<bool, IsDelayed>; + using DelayedTagCalc = std::true_type; + using ImmediateTagCalc = std::false_type; + + public: + + using RequestRef = std::unique_ptr<R>; + + protected: + + using Clock = std::chrono::steady_clock; + using TimePoint = Clock::time_point; + using Duration = std::chrono::milliseconds; + using MarkPoint = std::pair<TimePoint,Counter>; + + enum class ReadyOption {ignore, lowers, raises}; + + // forward decl for friend decls + template<double RequestTag::*, ReadyOption, bool> + struct ClientCompare; + + class ClientReq { + friend PriorityQueueBase; + + RequestTag tag; + C client_id; + RequestRef request; + + public: + + ClientReq(const RequestTag& _tag, + const C& _client_id, + RequestRef&& _request) : + tag(_tag), + client_id(_client_id), + request(std::move(_request)) + { + // empty + } + + friend std::ostream& operator<<(std::ostream& out, const ClientReq& c) { + out << "{ ClientReq:: tag:" << c.tag << " client:" << + c.client_id << " }"; + return out; + } + }; // class ClientReq + + struct RequestMeta { + C client_id; + RequestTag tag; + + RequestMeta(const C& _client_id, const RequestTag& _tag) : + client_id(_client_id), + tag(_tag) + { + // empty + } + }; + + public: + + // NOTE: ClientRec is in the "public" section for compatibility + // with g++ 4.8.4, which complains if it's not. By g++ 6.3.1 + // ClientRec could be "protected" with no issue. [See comments + // associated with function submit_top_request.] + class ClientRec { + friend PriorityQueueBase<C,R,IsDelayed,U1,B>; + + C client; + RequestTag prev_tag; + std::deque<ClientReq> requests; + + // amount added from the proportion tag as a result of + // an idle client becoming unidle + double prop_delta = 0.0; + + c::IndIntruHeapData reserv_heap_data {}; + c::IndIntruHeapData lim_heap_data {}; + c::IndIntruHeapData ready_heap_data {}; +#if USE_PROP_HEAP + c::IndIntruHeapData prop_heap_data {}; +#endif + + public: + + const ClientInfo* info; + bool idle; + Counter last_tick; + uint32_t cur_rho; + uint32_t cur_delta; + + ClientRec(C _client, + const ClientInfo* _info, + Counter current_tick) : + client(_client), + prev_tag(0.0, 0.0, 0.0, TimeZero), + info(_info), + idle(true), + last_tick(current_tick), + cur_rho(1), + cur_delta(1) + { + // empty + } + + inline const RequestTag& get_req_tag() const { + return prev_tag; + } + + static inline void assign_unpinned_tag(double& lhs, const double rhs) { + if (rhs != max_tag && rhs != min_tag) { + lhs = rhs; + } + } + + inline void update_req_tag(const RequestTag& _prev, + const Counter& _tick) { + assign_unpinned_tag(prev_tag.reservation, _prev.reservation); + assign_unpinned_tag(prev_tag.limit, _prev.limit); + assign_unpinned_tag(prev_tag.proportion, _prev.proportion); + prev_tag.arrival = _prev.arrival; + last_tick = _tick; + } + + inline void add_request(const RequestTag& tag, RequestRef&& request) { + requests.emplace_back(tag, client, std::move(request)); + } + + inline const ClientReq& next_request() const { + return requests.front(); + } + + inline ClientReq& next_request() { + return requests.front(); + } + + inline void pop_request() { + requests.pop_front(); + } + + inline bool has_request() const { + return !requests.empty(); + } + + inline size_t request_count() const { + return requests.size(); + } + + // NB: because a deque is the underlying structure, this + // operation might be expensive + bool remove_by_req_filter_fw(std::function<bool(RequestRef&&)> filter_accum) { + bool any_removed = false; + for (auto i = requests.begin(); + i != requests.end(); + /* no inc */) { + if (filter_accum(std::move(i->request))) { + any_removed = true; + i = requests.erase(i); + } else { + ++i; + } + } + return any_removed; + } + + // NB: because a deque is the underlying structure, this + // operation might be expensive + bool remove_by_req_filter_bw(std::function<bool(RequestRef&&)> filter_accum) { + bool any_removed = false; + for (auto i = requests.rbegin(); + i != requests.rend(); + /* no inc */) { + if (filter_accum(std::move(i->request))) { + any_removed = true; + i = decltype(i){ requests.erase(std::next(i).base()) }; + } else { + ++i; + } + } + return any_removed; + } + + inline bool + remove_by_req_filter(std::function<bool(RequestRef&&)> filter_accum, + bool visit_backwards) { + if (visit_backwards) { + return remove_by_req_filter_bw(filter_accum); + } else { + return remove_by_req_filter_fw(filter_accum); + } + } + + friend std::ostream& + operator<<(std::ostream& out, + const typename PriorityQueueBase::ClientRec& e) { + out << "{ ClientRec::" << + " client:" << e.client << + " prev_tag:" << e.prev_tag << + " req_count:" << e.requests.size() << + " top_req:"; + if (e.has_request()) { + out << e.next_request(); + } else { + out << "none"; + } + out << " }"; + + return out; + } + }; // class ClientRec + + using ClientRecRef = std::shared_ptr<ClientRec>; + + // when we try to get the next request, we'll be in one of three + // situations -- we'll have one to return, have one that can + // fire in the future, or not have any + enum class NextReqType { returning, future, none }; + + // specifies which queue next request will get popped from + enum class HeapId { reservation, ready }; + + // this is returned from next_req to tell the caller the situation + struct NextReq { + NextReqType type; + union { + HeapId heap_id; + Time when_ready; + }; + + inline explicit NextReq() : + type(NextReqType::none) + { } + + inline NextReq(HeapId _heap_id) : + type(NextReqType::returning), + heap_id(_heap_id) + { } + + inline NextReq(Time _when_ready) : + type(NextReqType::future), + when_ready(_when_ready) + { } + + // calls to this are clearer than calls to the default + // constructor + static inline NextReq none() { + return NextReq(); + } + }; + + + // a function that can be called to look up client information + using ClientInfoFunc = std::function<const ClientInfo*(const C&)>; + + + bool empty() const { + DataGuard g(data_mtx); + return (resv_heap.empty() || ! resv_heap.top().has_request()); + } + + + size_t client_count() const { + DataGuard g(data_mtx); + return resv_heap.size(); + } + + + size_t request_count() const { + DataGuard g(data_mtx); + size_t total = 0; + for (auto i = resv_heap.cbegin(); i != resv_heap.cend(); ++i) { + total += i->request_count(); + } + return total; + } + + + bool remove_by_req_filter(std::function<bool(RequestRef&&)> filter_accum, + bool visit_backwards = false) { + bool any_removed = false; + DataGuard g(data_mtx); + for (auto i : client_map) { + bool modified = + i.second->remove_by_req_filter(filter_accum, visit_backwards); + if (modified) { + resv_heap.adjust(*i.second); + limit_heap.adjust(*i.second); + ready_heap.adjust(*i.second); +#if USE_PROP_HEAP + prop_heap.adjust(*i.second); +#endif + any_removed = true; + } + } + return any_removed; + } + + + // use as a default value when no accumulator is provide + static void request_sink(RequestRef&& req) { + // do nothing + } + + + void remove_by_client(const C& client, + bool reverse = false, + std::function<void (RequestRef&&)> accum = request_sink) { + DataGuard g(data_mtx); + + auto i = client_map.find(client); + + if (i == client_map.end()) return; + + if (reverse) { + for (auto j = i->second->requests.rbegin(); + j != i->second->requests.rend(); + ++j) { + accum(std::move(j->request)); + } + } else { + for (auto j = i->second->requests.begin(); + j != i->second->requests.end(); + ++j) { + accum(std::move(j->request)); + } + } + + i->second->requests.clear(); + + resv_heap.adjust(*i->second); + limit_heap.adjust(*i->second); + ready_heap.adjust(*i->second); +#if USE_PROP_HEAP + prop_heap.adjust(*i->second); +#endif + } + + + unsigned get_heap_branching_factor() const { + return B; + } + + + void update_client_info(const C& client_id) { + DataGuard g(data_mtx); + auto client_it = client_map.find(client_id); + if (client_map.end() != client_it) { + ClientRec& client = (*client_it->second); + client.info = client_info_f(client_id); + } + } + + + void update_client_infos() { + DataGuard g(data_mtx); + for (auto i : client_map) { + i.second->info = client_info_f(i.second->client); + } + } + + + friend std::ostream& operator<<(std::ostream& out, + const PriorityQueueBase& q) { + std::lock_guard<decltype(q.data_mtx)> guard(q.data_mtx); + + out << "{ PriorityQueue::"; + for (const auto& c : q.client_map) { + out << " { client:" << c.first << ", record:" << *c.second << + " }"; + } + if (!q.resv_heap.empty()) { + const auto& resv = q.resv_heap.top(); + out << " { reservation_top:" << resv << " }"; + const auto& ready = q.ready_heap.top(); + out << " { ready_top:" << ready << " }"; + const auto& limit = q.limit_heap.top(); + out << " { limit_top:" << limit << " }"; + } else { + out << " HEAPS-EMPTY"; + } + out << " }"; + + return out; + } + + // for debugging + void display_queues(std::ostream& out, + bool show_res = true, + bool show_lim = true, + bool show_ready = true, + bool show_prop = true) const { + auto filter = [](const ClientRec& e)->bool { return true; }; + DataGuard g(data_mtx); + if (show_res) { + resv_heap.display_sorted(out << "RESER:", filter); + } + if (show_lim) { + limit_heap.display_sorted(out << "LIMIT:", filter); + } + if (show_ready) { + ready_heap.display_sorted(out << "READY:", filter); + } +#if USE_PROP_HEAP + if (show_prop) { + prop_heap.display_sorted(out << "PROPO:", filter); + } +#endif + } // display_queues + + + protected: + + // The ClientCompare functor is essentially doing a precedes? + // operator, returning true if and only if the first parameter + // must precede the second parameter. If the second must precede + // the first, or if they are equivalent, false should be + // returned. The reason for this behavior is that it will be + // called to test if two items are out of order and if true is + // returned it will reverse the items. Therefore false is the + // default return when it doesn't matter to prevent unnecessary + // re-ordering. + // + // The template is supporting variations in sorting based on the + // heap in question and allowing these variations to be handled + // at compile-time. + // + // tag_field determines which tag is being used for comparison + // + // ready_opt determines how the ready flag influences the sort + // + // use_prop_delta determines whether the proportional delta is + // added in for comparison + template<double RequestTag::*tag_field, + ReadyOption ready_opt, + bool use_prop_delta> + struct ClientCompare { + bool operator()(const ClientRec& n1, const ClientRec& n2) const { + if (n1.has_request()) { + if (n2.has_request()) { + const auto& t1 = n1.next_request().tag; + const auto& t2 = n2.next_request().tag; + if (ReadyOption::ignore == ready_opt || t1.ready == t2.ready) { + // if we don't care about ready or the ready values are the same + if (use_prop_delta) { + return (t1.*tag_field + n1.prop_delta) < + (t2.*tag_field + n2.prop_delta); + } else { + return t1.*tag_field < t2.*tag_field; + } + } else if (ReadyOption::raises == ready_opt) { + // use_ready == true && the ready fields are different + return t1.ready; + } else { + return t2.ready; + } + } else { + // n1 has request but n2 does not + return true; + } + } else if (n2.has_request()) { + // n2 has request but n1 does not + return false; + } else { + // both have none; keep stable w false + return false; + } + } + }; + + ClientInfoFunc client_info_f; + static constexpr bool is_dynamic_cli_info_f = U1; + + mutable std::mutex data_mtx; + using DataGuard = std::lock_guard<decltype(data_mtx)>; + + // stable mapping between client ids and client queues + std::map<C,ClientRecRef> client_map; + + c::IndIntruHeap<ClientRecRef, + ClientRec, + &ClientRec::reserv_heap_data, + ClientCompare<&RequestTag::reservation, + ReadyOption::ignore, + false>, + B> resv_heap; +#if USE_PROP_HEAP + c::IndIntruHeap<ClientRecRef, + ClientRec, + &ClientRec::prop_heap_data, + ClientCompare<&RequestTag::proportion, + ReadyOption::ignore, + true>, + B> prop_heap; +#endif + c::IndIntruHeap<ClientRecRef, + ClientRec, + &ClientRec::lim_heap_data, + ClientCompare<&RequestTag::limit, + ReadyOption::lowers, + false>, + B> limit_heap; + c::IndIntruHeap<ClientRecRef, + ClientRec, + &ClientRec::ready_heap_data, + ClientCompare<&RequestTag::proportion, + ReadyOption::raises, + true>, + B> ready_heap; + + AtLimit at_limit; + RejectThreshold reject_threshold = 0; + + double anticipation_timeout; + + std::atomic_bool finishing; + + // every request creates a tick + Counter tick = 0; + + // performance data collection + size_t reserv_sched_count = 0; + size_t prop_sched_count = 0; + size_t limit_break_sched_count = 0; + + Duration idle_age; + Duration erase_age; + Duration check_time; + std::deque<MarkPoint> clean_mark_points; + // max number of clients to erase at a time + Counter erase_max; + // unfinished last erase point + Counter last_erase_point = 0; + + // NB: All threads declared at end, so they're destructed first! + + std::unique_ptr<RunEvery> cleaning_job; + + // helper function to return the value of a variant if it matches the + // given type T, or a default value of T otherwise + template <typename T, typename Variant> + static T get_or_default(const Variant& param, T default_value) { + const T *p = boost::get<T>(¶m); + return p ? *p : default_value; + } + + // COMMON constructor that others feed into; we can accept three + // different variations of durations + template<typename Rep, typename Per> + PriorityQueueBase(ClientInfoFunc _client_info_f, + std::chrono::duration<Rep,Per> _idle_age, + std::chrono::duration<Rep,Per> _erase_age, + std::chrono::duration<Rep,Per> _check_time, + AtLimitParam at_limit_param, + double _anticipation_timeout) : + client_info_f(_client_info_f), + at_limit(get_or_default(at_limit_param, AtLimit::Reject)), + reject_threshold(get_or_default(at_limit_param, RejectThreshold{0})), + anticipation_timeout(_anticipation_timeout), + finishing(false), + idle_age(std::chrono::duration_cast<Duration>(_idle_age)), + erase_age(std::chrono::duration_cast<Duration>(_erase_age)), + check_time(std::chrono::duration_cast<Duration>(_check_time)), + erase_max(standard_erase_max) + { + assert(_erase_age >= _idle_age); + assert(_check_time < _idle_age); + // AtLimit::Reject depends on ImmediateTagCalc + assert(at_limit != AtLimit::Reject || !IsDelayed); + cleaning_job = + std::unique_ptr<RunEvery>( + new RunEvery(check_time, + std::bind(&PriorityQueueBase::do_clean, this))); + } + + + ~PriorityQueueBase() { + finishing = true; + } + + + inline const ClientInfo* get_cli_info(ClientRec& client) const { + if (is_dynamic_cli_info_f) { + client.info = client_info_f(client.client); + } + return client.info; + } + + // data_mtx must be held by caller + RequestTag initial_tag(DelayedTagCalc delayed, ClientRec& client, + const ReqParams& params, Time time, Cost cost) { + RequestTag tag(0, 0, 0, time, 0, 0, cost); + + // only calculate a tag if the request is going straight to the front + if (!client.has_request()) { + const ClientInfo* client_info = get_cli_info(client); + assert(client_info); + tag = RequestTag(client.get_req_tag(), *client_info, + params, time, cost, anticipation_timeout); + + // copy tag to previous tag for client + client.update_req_tag(tag, tick); + } + return tag; + } + + // data_mtx must be held by caller + RequestTag initial_tag(ImmediateTagCalc imm, ClientRec& client, + const ReqParams& params, Time time, Cost cost) { + // calculate the tag unconditionally + const ClientInfo* client_info = get_cli_info(client); + assert(client_info); + RequestTag tag(client.get_req_tag(), *client_info, + params, time, cost, anticipation_timeout); + + // copy tag to previous tag for client + client.update_req_tag(tag, tick); + return tag; + } + + // data_mtx must be held by caller. returns 0 on success. when using + // AtLimit::Reject, requests that would exceed their limit are rejected + // with EAGAIN, and the queue will not take ownership of the given + // 'request' argument + int do_add_request(RequestRef&& request, + const C& client_id, + const ReqParams& req_params, + const Time time, + const Cost cost = 1u) { + ++tick; + + auto insert = client_map.emplace(client_id, ClientRecRef{}); + if (insert.second) { + // new client entry + const ClientInfo* info = client_info_f(client_id); + auto client_rec = std::make_shared<ClientRec>(client_id, info, tick); + resv_heap.push(client_rec); +#if USE_PROP_HEAP + prop_heap.push(client_rec); +#endif + limit_heap.push(client_rec); + ready_heap.push(client_rec); + insert.first->second = std::move(client_rec); + } + + // for convenience, we'll create a reference to the shared pointer + ClientRec& client = *insert.first->second; + + if (client.idle) { + // We need to do an adjustment so that idle clients compete + // fairly on proportional tags since those tags may have + // drifted from real-time. Either use the lowest existing + // proportion tag -- O(1) -- or the client with the lowest + // previous proportion tag -- O(n) where n = # clients. + // + // So we don't have to maintain a proportional queue that + // keeps the minimum on proportional tag alone (we're + // instead using a ready queue), we'll have to check each + // client. + // + // The alternative would be to maintain a proportional queue + // (define USE_PROP_TAG) and do an O(1) operation here. + + // Was unable to confirm whether equality testing on + // std::numeric_limits<double>::max() is guaranteed, so + // we'll use a compile-time calculated trigger that is one + // third the max, which should be much larger than any + // expected organic value. + constexpr double lowest_prop_tag_trigger = + std::numeric_limits<double>::max() / 3.0; + + double lowest_prop_tag = std::numeric_limits<double>::max(); + for (auto const &c : client_map) { + // don't use ourselves (or anything else that might be + // listed as idle) since we're now in the map + if (!c.second->idle) { + double p; + // use either lowest proportion tag or previous proportion tag + if (c.second->has_request()) { + p = c.second->next_request().tag.proportion + + c.second->prop_delta; + } else { + p = c.second->get_req_tag().proportion + c.second->prop_delta; + } + + if (p < lowest_prop_tag) { + lowest_prop_tag = p; + } + } + } + + // if this conditional does not fire, it + if (lowest_prop_tag < lowest_prop_tag_trigger) { + client.prop_delta = lowest_prop_tag - time; + } + client.idle = false; + } // if this client was idle + + RequestTag tag = initial_tag(TagCalc{}, client, req_params, time, cost); + + if (at_limit == AtLimit::Reject && + tag.limit > time + reject_threshold) { + // if the client is over its limit, reject it here + return EAGAIN; + } + + client.add_request(tag, std::move(request)); + if (1 == client.requests.size()) { + // NB: can the following 4 calls to adjust be changed + // promote? Can adding a request ever demote a client in the + // heaps? + resv_heap.adjust(client); + limit_heap.adjust(client); + ready_heap.adjust(client); +#if USE_PROP_HEAP + prop_heap.adjust(client); +#endif + } + + client.cur_rho = req_params.rho; + client.cur_delta = req_params.delta; + + resv_heap.adjust(client); + limit_heap.adjust(client); + ready_heap.adjust(client); +#if USE_PROP_HEAP + prop_heap.adjust(client); +#endif + return 0; + } // do_add_request + + // data_mtx must be held by caller + void update_next_tag(DelayedTagCalc delayed, ClientRec& top, + const RequestTag& tag) { + if (top.has_request()) { + // perform delayed tag calculation on the next request + ClientReq& next_first = top.next_request(); + const ClientInfo* client_info = get_cli_info(top); + assert(client_info); + next_first.tag = RequestTag(tag, *client_info, + top.cur_delta, top.cur_rho, + next_first.tag.arrival, + next_first.tag.cost, + anticipation_timeout); + // copy tag to previous tag for client + top.update_req_tag(next_first.tag, tick); + } + } + + void update_next_tag(ImmediateTagCalc imm, ClientRec& top, + const RequestTag& tag) { + // the next tag was already calculated on insertion + } + + // data_mtx should be held when called; top of heap should have + // a ready request + template<typename C1, IndIntruHeapData ClientRec::*C2, typename C3> + RequestTag pop_process_request(IndIntruHeap<C1, ClientRec, C2, C3, B>& heap, + std::function<void(const C& client, + const Cost cost, + RequestRef& request)> process) { + // gain access to data + ClientRec& top = heap.top(); + + Cost request_cost = top.next_request().tag.cost; + RequestRef request = std::move(top.next_request().request); + RequestTag tag = top.next_request().tag; + + // pop request and adjust heaps + top.pop_request(); + + update_next_tag(TagCalc{}, top, tag); + + resv_heap.demote(top); + limit_heap.adjust(top); +#if USE_PROP_HEAP + prop_heap.demote(top); +#endif + ready_heap.demote(top); + + // process + process(top.client, request_cost, request); + + return tag; + } // pop_process_request + + + // data_mtx must be held by caller + void reduce_reservation_tags(DelayedTagCalc delayed, ClientRec& client, + const RequestTag& tag) { + if (!client.requests.empty()) { + // only maintain a tag for the first request + auto& r = client.requests.front(); + r.tag.reservation -= + client.info->reservation_inv * std::max(uint32_t(1), tag.rho); + } + } + + // data_mtx should be held when called + void reduce_reservation_tags(ImmediateTagCalc imm, ClientRec& client, + const RequestTag& tag) { + double res_offset = + client.info->reservation_inv * std::max(uint32_t(1), tag.rho); + for (auto& r : client.requests) { + r.tag.reservation -= res_offset; + } + } + + // data_mtx should be held when called + void reduce_reservation_tags(const C& client_id, const RequestTag& tag) { + auto client_it = client_map.find(client_id); + + // means the client was cleaned from map; should never happen + // as long as cleaning times are long enough + assert(client_map.end() != client_it); + ClientRec& client = *client_it->second; + reduce_reservation_tags(TagCalc{}, client, tag); + + // don't forget to update previous tag + client.prev_tag.reservation -= + client.info->reservation_inv * std::max(uint32_t(1), tag.rho); + resv_heap.promote(client); + } + + + // data_mtx should be held when called + NextReq do_next_request(Time now) { + // if reservation queue is empty, all are empty (i.e., no + // active clients) + if(resv_heap.empty()) { + return NextReq::none(); + } + + // try constraint (reservation) based scheduling + + auto& reserv = resv_heap.top(); + if (reserv.has_request() && + reserv.next_request().tag.reservation <= now) { + return NextReq(HeapId::reservation); + } + + // no existing reservations before now, so try weight-based + // scheduling + + // all items that are within limit are eligible based on + // priority + auto limits = &limit_heap.top(); + while (limits->has_request() && + !limits->next_request().tag.ready && + limits->next_request().tag.limit <= now) { + limits->next_request().tag.ready = true; + ready_heap.promote(*limits); + limit_heap.demote(*limits); + + limits = &limit_heap.top(); + } + + auto& readys = ready_heap.top(); + if (readys.has_request() && + readys.next_request().tag.ready && + readys.next_request().tag.proportion < max_tag) { + return NextReq(HeapId::ready); + } + + // if nothing is schedulable by reservation or + // proportion/weight, and if we allow limit break, try to + // schedule something with the lowest proportion tag or + // alternatively lowest reservation tag. + if (at_limit == AtLimit::Allow) { + if (readys.has_request() && + readys.next_request().tag.proportion < max_tag) { + return NextReq(HeapId::ready); + } else if (reserv.has_request() && + reserv.next_request().tag.reservation < max_tag) { + return NextReq(HeapId::reservation); + } + } + + // nothing scheduled; make sure we re-run when next + // reservation item or next limited item comes up + + Time next_call = TimeMax; + if (resv_heap.top().has_request()) { + next_call = + min_not_0_time(next_call, + resv_heap.top().next_request().tag.reservation); + } + if (limit_heap.top().has_request()) { + const auto& next = limit_heap.top().next_request(); + assert(!next.tag.ready || max_tag == next.tag.proportion); + next_call = min_not_0_time(next_call, next.tag.limit); + } + if (next_call < TimeMax) { + return NextReq(next_call); + } else { + return NextReq::none(); + } + } // do_next_request + + + // if possible is not zero and less than current then return it; + // otherwise return current; the idea is we're trying to find + // the minimal time but ignoring zero + static inline const Time& min_not_0_time(const Time& current, + const Time& possible) { + return TimeZero == possible ? current : std::min(current, possible); + } + + + /* + * This is being called regularly by RunEvery. Every time it's + * called it notes the time and delta counter (mark point) in a + * deque. It also looks at the deque to find the most recent + * mark point that is older than clean_age. It then walks the + * map and delete all server entries that were last used before + * that mark point. + */ + void do_clean() { + TimePoint now = std::chrono::steady_clock::now(); + DataGuard g(data_mtx); + clean_mark_points.emplace_back(MarkPoint(now, tick)); + + // first erase the super-old client records + + Counter erase_point = last_erase_point; + auto point = clean_mark_points.front(); + while (point.first <= now - erase_age) { + last_erase_point = point.second; + erase_point = last_erase_point; + clean_mark_points.pop_front(); + point = clean_mark_points.front(); + } + + Counter idle_point = 0; + for (auto i : clean_mark_points) { + if (i.first <= now - idle_age) { + idle_point = i.second; + } else { + break; + } + } + + Counter erased_num = 0; + if (erase_point > 0 || idle_point > 0) { + for (auto i = client_map.begin(); i != client_map.end(); /* empty */) { + auto i2 = i++; + if (erase_point && + erased_num < erase_max && + i2->second->last_tick <= erase_point) { + delete_from_heaps(i2->second); + client_map.erase(i2); + erased_num++; + } else if (idle_point && i2->second->last_tick <= idle_point) { + i2->second->idle = true; + } + } // for + + auto wperiod = check_time; + if (erased_num >= erase_max) { + wperiod = duration_cast<milliseconds>(aggressive_check_time); + } else { + // clean finished, refresh + last_erase_point = 0; + } + cleaning_job->try_update(wperiod); + } // if + } // do_clean + + + // data_mtx must be held by caller + template<IndIntruHeapData ClientRec::*C1,typename C2> + void delete_from_heap(ClientRecRef& client, + c::IndIntruHeap<ClientRecRef,ClientRec,C1,C2,B>& heap) { + auto i = heap.at(client); + heap.remove(i); + } + + + // data_mtx must be held by caller + void delete_from_heaps(ClientRecRef& client) { + delete_from_heap(client, resv_heap); +#if USE_PROP_HEAP + delete_from_heap(client, prop_heap); +#endif + delete_from_heap(client, limit_heap); + delete_from_heap(client, ready_heap); + } + }; // class PriorityQueueBase + + + template<typename C, typename R, bool IsDelayed=false, bool U1=false, unsigned B=2> + class PullPriorityQueue : public PriorityQueueBase<C,R,IsDelayed,U1,B> { + using super = PriorityQueueBase<C,R,IsDelayed,U1,B>; + + public: + + // When a request is pulled, this is the return type. + struct PullReq { + struct Retn { + C client; + typename super::RequestRef request; + PhaseType phase; + Cost cost; + }; + + typename super::NextReqType type; + boost::variant<Retn,Time> data; + + bool is_none() const { return type == super::NextReqType::none; } + + bool is_retn() const { return type == super::NextReqType::returning; } + Retn& get_retn() { + return boost::get<Retn>(data); + } + + bool is_future() const { return type == super::NextReqType::future; } + Time getTime() const { return boost::get<Time>(data); } + }; + + +#ifdef PROFILE + ProfileTimer<std::chrono::nanoseconds> pull_request_timer; + ProfileTimer<std::chrono::nanoseconds> add_request_timer; +#endif + + template<typename Rep, typename Per> + PullPriorityQueue(typename super::ClientInfoFunc _client_info_f, + std::chrono::duration<Rep,Per> _idle_age, + std::chrono::duration<Rep,Per> _erase_age, + std::chrono::duration<Rep,Per> _check_time, + AtLimitParam at_limit_param = AtLimit::Wait, + double _anticipation_timeout = 0.0) : + super(_client_info_f, + _idle_age, _erase_age, _check_time, + at_limit_param, _anticipation_timeout) + { + // empty + } + + + // pull convenience constructor + PullPriorityQueue(typename super::ClientInfoFunc _client_info_f, + AtLimitParam at_limit_param = AtLimit::Wait, + double _anticipation_timeout = 0.0) : + PullPriorityQueue(_client_info_f, + standard_idle_age, + standard_erase_age, + standard_check_time, + at_limit_param, + _anticipation_timeout) + { + // empty + } + + + int add_request(R&& request, + const C& client_id, + const ReqParams& req_params, + const Cost cost = 1u) { + return add_request(typename super::RequestRef(new R(std::move(request))), + client_id, + req_params, + get_time(), + cost); + } + + + int add_request(R&& request, + const C& client_id, + const Cost cost = 1u) { + static const ReqParams null_req_params; + return add_request(typename super::RequestRef(new R(std::move(request))), + client_id, + null_req_params, + get_time(), + cost); + } + + + int add_request_time(R&& request, + const C& client_id, + const ReqParams& req_params, + const Time time, + const Cost cost = 1u) { + return add_request(typename super::RequestRef(new R(std::move(request))), + client_id, + req_params, + time, + cost); + } + + + int add_request(typename super::RequestRef&& request, + const C& client_id, + const ReqParams& req_params, + const Cost cost = 1u) { + return add_request(request, req_params, client_id, get_time(), cost); + } + + + int add_request(typename super::RequestRef&& request, + const C& client_id, + const Cost cost = 1u) { + static const ReqParams null_req_params; + return add_request(request, null_req_params, client_id, get_time(), cost); + } + + + // this does the work; the versions above provide alternate interfaces + int add_request(typename super::RequestRef&& request, + const C& client_id, + const ReqParams& req_params, + const Time time, + const Cost cost = 1u) { + typename super::DataGuard g(this->data_mtx); +#ifdef PROFILE + add_request_timer.start(); +#endif + int r = super::do_add_request(std::move(request), + client_id, + req_params, + time, + cost); + // no call to schedule_request for pull version +#ifdef PROFILE + add_request_timer.stop(); +#endif + return r; + } + + + inline PullReq pull_request() { + return pull_request(get_time()); + } + + + PullReq pull_request(const Time now) { + PullReq result; + typename super::DataGuard g(this->data_mtx); +#ifdef PROFILE + pull_request_timer.start(); +#endif + + typename super::NextReq next = super::do_next_request(now); + result.type = next.type; + switch(next.type) { + case super::NextReqType::none: + return result; + case super::NextReqType::future: + result.data = next.when_ready; + return result; + case super::NextReqType::returning: + // to avoid nesting, break out and let code below handle this case + break; + default: + assert(false); + } + + // we'll only get here if we're returning an entry + + auto process_f = + [&] (PullReq& pull_result, PhaseType phase) -> + std::function<void(const C&, + uint64_t, + typename super::RequestRef&)> { + return [&pull_result, phase](const C& client, + const Cost request_cost, + typename super::RequestRef& request) { + pull_result.data = typename PullReq::Retn{ client, + std::move(request), + phase, + request_cost }; + }; + }; + + switch(next.heap_id) { + case super::HeapId::reservation: + (void) super::pop_process_request(this->resv_heap, + process_f(result, + PhaseType::reservation)); + ++this->reserv_sched_count; + break; + case super::HeapId::ready: + { + auto tag = super::pop_process_request(this->ready_heap, + process_f(result, PhaseType::priority)); + // need to use retn temporarily + auto& retn = boost::get<typename PullReq::Retn>(result.data); + super::reduce_reservation_tags(retn.client, tag); + } + ++this->prop_sched_count; + break; + default: + assert(false); + } + +#ifdef PROFILE + pull_request_timer.stop(); +#endif + return result; + } // pull_request + + + protected: + + + // data_mtx should be held when called; unfortunately this + // function has to be repeated in both push & pull + // specializations + typename super::NextReq next_request() { + return next_request(get_time()); + } + }; // class PullPriorityQueue + + + // PUSH version + template<typename C, typename R, bool IsDelayed=false, bool U1=false, unsigned B=2> + class PushPriorityQueue : public PriorityQueueBase<C,R,IsDelayed,U1,B> { + + protected: + + using super = PriorityQueueBase<C,R,IsDelayed,U1,B>; + + public: + + // a function to see whether the server can handle another request + using CanHandleRequestFunc = std::function<bool(void)>; + + // a function to submit a request to the server; the second + // parameter is a callback when it's completed + using HandleRequestFunc = + std::function<void(const C&,typename super::RequestRef,PhaseType,uint64_t)>; + + protected: + + CanHandleRequestFunc can_handle_f; + HandleRequestFunc handle_f; + // for handling timed scheduling + std::mutex sched_ahead_mtx; + std::condition_variable sched_ahead_cv; + Time sched_ahead_when = TimeZero; + +#ifdef PROFILE + public: + ProfileTimer<std::chrono::nanoseconds> add_request_timer; + ProfileTimer<std::chrono::nanoseconds> request_complete_timer; + protected: +#endif + + // NB: threads declared last, so constructed last and destructed first + + std::thread sched_ahead_thd; + + public: + + // push full constructor + template<typename Rep, typename Per> + PushPriorityQueue(typename super::ClientInfoFunc _client_info_f, + CanHandleRequestFunc _can_handle_f, + HandleRequestFunc _handle_f, + std::chrono::duration<Rep,Per> _idle_age, + std::chrono::duration<Rep,Per> _erase_age, + std::chrono::duration<Rep,Per> _check_time, + AtLimitParam at_limit_param = AtLimit::Wait, + double anticipation_timeout = 0.0) : + super(_client_info_f, + _idle_age, _erase_age, _check_time, + at_limit_param, anticipation_timeout) + { + can_handle_f = _can_handle_f; + handle_f = _handle_f; + sched_ahead_thd = std::thread(&PushPriorityQueue::run_sched_ahead, this); + } + + + // push convenience constructor + PushPriorityQueue(typename super::ClientInfoFunc _client_info_f, + CanHandleRequestFunc _can_handle_f, + HandleRequestFunc _handle_f, + AtLimitParam at_limit_param = AtLimit::Wait, + double _anticipation_timeout = 0.0) : + PushPriorityQueue(_client_info_f, + _can_handle_f, + _handle_f, + standard_idle_age, + standard_erase_age, + standard_check_time, + at_limit_param, + _anticipation_timeout) + { + // empty + } + + + ~PushPriorityQueue() { + this->finishing = true; + { + std::lock_guard<std::mutex> l(sched_ahead_mtx); + sched_ahead_cv.notify_one(); + } + sched_ahead_thd.join(); + } + + public: + + int add_request(R&& request, + const C& client_id, + const ReqParams& req_params, + const Cost cost = 1u) { + return add_request(typename super::RequestRef(new R(std::move(request))), + client_id, + req_params, + get_time(), + cost); + } + + + int add_request(typename super::RequestRef&& request, + const C& client_id, + const ReqParams& req_params, + const Cost cost = 1u) { + return add_request(request, req_params, client_id, get_time(), cost); + } + + + int add_request_time(const R& request, + const C& client_id, + const ReqParams& req_params, + const Time time, + const Cost cost = 1u) { + return add_request(typename super::RequestRef(new R(request)), + client_id, + req_params, + time, + cost); + } + + + int add_request(typename super::RequestRef&& request, + const C& client_id, + const ReqParams& req_params, + const Time time, + const Cost cost = 1u) { + typename super::DataGuard g(this->data_mtx); +#ifdef PROFILE + add_request_timer.start(); +#endif + int r = super::do_add_request(std::move(request), + client_id, + req_params, + time, + cost); + if (r == 0) { + schedule_request(); + } +#ifdef PROFILE + add_request_timer.stop(); +#endif + return r; + } + + + void request_completed() { + typename super::DataGuard g(this->data_mtx); +#ifdef PROFILE + request_complete_timer.start(); +#endif + schedule_request(); +#ifdef PROFILE + request_complete_timer.stop(); +#endif + } + + protected: + + // data_mtx should be held when called; furthermore, the heap + // should not be empty and the top element of the heap should + // not be already handled + // + // NOTE: the use of "super::ClientRec" in either the template + // construct or as a parameter to submit_top_request generated + // a compiler error in g++ 4.8.4, when ClientRec was + // "protected" rather than "public". By g++ 6.3.1 this was not + // an issue. But for backwards compatibility + // PriorityQueueBase::ClientRec is public. + template<typename C1, + IndIntruHeapData super::ClientRec::*C2, + typename C3, + unsigned B4> + typename super::RequestMeta + submit_top_request(IndIntruHeap<C1,typename super::ClientRec,C2,C3,B4>& heap, + PhaseType phase) { + C client_result; + RequestTag tag = super::pop_process_request(heap, + [this, phase, &client_result] + (const C& client, + const Cost request_cost, + typename super::RequestRef& request) { + client_result = client; + handle_f(client, std::move(request), phase, request_cost); + }); + typename super::RequestMeta req(client_result, tag); + return req; + } + + + // data_mtx should be held when called + void submit_request(typename super::HeapId heap_id) { + switch(heap_id) { + case super::HeapId::reservation: + // don't need to note client + (void) submit_top_request(this->resv_heap, PhaseType::reservation); + // unlike the other two cases, we do not reduce reservation + // tags here + ++this->reserv_sched_count; + break; + case super::HeapId::ready: + { + auto req = submit_top_request(this->ready_heap, PhaseType::priority); + super::reduce_reservation_tags(req.client_id, req.tag); + } + ++this->prop_sched_count; + break; + default: + assert(false); + } + } // submit_request + + + // data_mtx should be held when called; unfortunately this + // function has to be repeated in both push & pull + // specializations + typename super::NextReq next_request() { + return next_request(get_time()); + } + + + // data_mtx should be held when called; overrides member + // function in base class to add check for whether a request can + // be pushed to the server + typename super::NextReq next_request(Time now) { + if (!can_handle_f()) { + typename super::NextReq result; + result.type = super::NextReqType::none; + return result; + } else { + return super::do_next_request(now); + } + } // next_request + + + // data_mtx should be held when called + void schedule_request() { + typename super::NextReq next_req = next_request(); + switch (next_req.type) { + case super::NextReqType::none: + return; + case super::NextReqType::future: + sched_at(next_req.when_ready); + break; + case super::NextReqType::returning: + submit_request(next_req.heap_id); + break; + default: + assert(false); + } + } + + + // this is the thread that handles running schedule_request at + // future times when nothing can be scheduled immediately + void run_sched_ahead() { + std::unique_lock<std::mutex> l(sched_ahead_mtx); + + while (!this->finishing) { + // predicate for cond.wait() + const auto pred = [this] () -> bool { return this->finishing; }; + + if (TimeZero == sched_ahead_when) { + sched_ahead_cv.wait(l, pred); + } else { + // cast from Time -> duration<Time> -> Duration -> TimePoint + const auto until = typename super::TimePoint{ + duration_cast<typename super::Duration>( + std::chrono::duration<Time>{sched_ahead_when})}; + sched_ahead_cv.wait_until(l, until, pred); + sched_ahead_when = TimeZero; + if (this->finishing) return; + + l.unlock(); + if (!this->finishing) { + typename super::DataGuard g(this->data_mtx); + schedule_request(); + } + l.lock(); + } + } + } + + + void sched_at(Time when) { + std::lock_guard<std::mutex> l(sched_ahead_mtx); + if (this->finishing) return; + if (TimeZero == sched_ahead_when || when < sched_ahead_when) { + sched_ahead_when = when; + sched_ahead_cv.notify_one(); + } + } + }; // class PushPriorityQueue + + } // namespace dmclock +} // namespace crimson diff --git a/src/dmclock/src/dmclock_util.cc b/src/dmclock/src/dmclock_util.cc new file mode 100644 index 000000000..75960ed74 --- /dev/null +++ b/src/dmclock/src/dmclock_util.cc @@ -0,0 +1,34 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2017 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include <signal.h> + +#include <iomanip> +#include <sstream> + +#include "dmclock_util.h" + + +std::string crimson::dmclock::format_time(const Time& time, unsigned modulo) { + long subtract = long(time / modulo) * modulo; + std::stringstream ss; + ss << std::fixed << std::setprecision(4) << (time - subtract); + return ss.str(); +} + + +void crimson::dmclock::debugger() { + raise(SIGCONT); +} diff --git a/src/dmclock/src/dmclock_util.h b/src/dmclock/src/dmclock_util.h new file mode 100644 index 000000000..f7e561ff7 --- /dev/null +++ b/src/dmclock/src/dmclock_util.h @@ -0,0 +1,60 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2017 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <unistd.h> +#include <assert.h> +#include <sys/time.h> + +#include <limits> +#include <cmath> +#include <chrono> + + +namespace crimson { + namespace dmclock { + // we're using double to represent time, but we could change it by + // changing the following declarations (and by making sure a min + // function existed) + using Time = double; + static const Time TimeZero = 0.0; + static const Time TimeMax = std::numeric_limits<Time>::max(); + static const double NaN = nan(""); + + + inline Time get_time() { +#if defined(__linux__) + struct timespec now; + auto result = clock_gettime(CLOCK_REALTIME, &now); + (void) result; // reference result in case assert is compiled out + assert(0 == result); + return now.tv_sec + (now.tv_nsec / 1.0e9); +#else + struct timeval now; + auto result = gettimeofday(&now, NULL); + (void) result; // reference result in case assert is compiled out + assert(0 == result); + return now.tv_sec + (now.tv_usec / 1.0e6); +#endif + } + + std::string format_time(const Time& time, unsigned modulo = 1000); + + void debugger(); + + } // namespace dmclock +} // namespace crimson diff --git a/src/dmclock/support/src/debug.h b/src/dmclock/support/src/debug.h new file mode 100644 index 000000000..d8e6713fd --- /dev/null +++ b/src/dmclock/support/src/debug.h @@ -0,0 +1,24 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <signal.h> + + +inline void debugger() { + raise(SIGCONT); +} diff --git a/src/dmclock/support/src/heap.h b/src/dmclock/support/src/heap.h new file mode 100644 index 000000000..6a1f9963a --- /dev/null +++ b/src/dmclock/support/src/heap.h @@ -0,0 +1,247 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <vector> +#include <ostream> + +#include "assert.h" + + +namespace crimson { + + /* + * T : type of data held in the heap. + * + * C : class that implements operator() with two arguments and + * returns a boolean when the first argument is greater than (higher + * in priority than) the second. + */ + template<typename T, typename C> + class Heap { + + public: + + class iterator { + + friend Heap<T,C>; + + Heap<T,C>& heap; + int index; + + iterator(Heap<T,C>& _heap, int _index) : + heap(_heap), + index(_index) + { + // empty + } + + public: + + iterator(iterator&& other) : + heap(other.heap), + index(other.index) + { + // empty + } + + iterator& operator++() { + ++index; + return *this; + } + + bool operator==(const iterator& other) const { + return index == other.index; + } + + bool operator!=(const iterator& other) const { + return !(*this == other); + } + + T& operator*() { + return heap.data[index]; + } + + // the item this iterator refers to + void increase() { + heap.siftUp(index); + } + }; // class iterator + + friend iterator; + + protected: + + std::vector<T> data; + int count; + C comparator; + + // parent(0) should be a negative value, which it is due to + // truncating towards negative infinity + static inline int parent(int i) { return (i - 1) / 2; } + + static inline int lhs(int i) { return 2*i + 1; } + + static inline int rhs(int i) { return 2*i + 2; } + + void siftUp(int i) { + assert(i < count); + + while (i > 0) { + int pi = parent(i); + if (!comparator(data[i], data[pi])) { + break; + } + + std::swap(data[i], data[pi]); + i = pi; + } + } + + void siftDown(int i) { + while (i < count) { + int li = lhs(i); + int ri = rhs(i); + + if (li < count) { + if (comparator(data[li], data[i])) { + if (ri < count && comparator(data[ri], data[li])) { + std::swap(data[i], data[ri]); + i = ri; + } else { + std::swap(data[i], data[li]); + i = li; + } + } else if (ri < count && comparator(data[ri], data[i])) { + std::swap(data[i], data[ri]); + i = ri; + } else { + break; + } + } else { + break; + } + } + } + + + public: + + Heap() : + count(0) + { + // empty + } + + Heap(const Heap<T,C>& other) { + data.resize(other.data.size()); + for (int i = 0; i < other.count; ++i) { + data[i] = other.data[i]; + } + count = other.count; + } + + const Heap<T,C>& operator=(const Heap<T,C>& other) { + data.resize(other.data.size()); + for (int i = 0; i < other.count; ++i) { + data[i] = other.data[i]; + } + count = other.count; + return *this; + } + + bool empty() const { return 0 == count; } + + T& top() { return data[0]; } + + void push(T item) { + int i = count++; + data.push_back(item); + siftUp(i); + } + + void pop() { + data[0] = data[--count]; + data.resize(count); + siftDown(0); + } + + void updateTop() { + siftDown(0); + } + + void clear() { + count = 0; + data.resize(0); + } + + iterator begin() { + return iterator(*this, 0); + } + + iterator end() { + return iterator(*this, count); + } + + std::ostream& displaySorted(std::ostream& out, + std::function<bool(const T&)> filter, + bool insert_line_breaks = true) const { + Heap<T,C> temp = *this; + + bool first = true; + out << "[ "; + + while(!temp.empty()) { + const T& top = temp.top(); + if (filter(top)) { + if (!first) { + out << ", "; + } + if (insert_line_breaks) { + out << std::endl << " "; + } + out << temp.top(); + first = false; + } + temp.pop(); + } + + out << " ]"; + if (insert_line_breaks) { + out << std::endl; + } + return out; + } + + template<typename T1, typename T2> + friend std::ostream& operator<<(std::ostream&, const Heap<T1,T2>&); + }; // class Heap + + + template<typename T1, typename T2> + std::ostream& operator<<(std::ostream& out, const Heap<T1,T2>& h) { + out << "[ "; + if (h.count) { + out << h.data[0]; + } + for (int i = 1; i < h.count; i++) { + out << ", " << h.data[i]; + } + out << " ]"; + return out; + } +} // namespace diff --git a/src/dmclock/support/src/indirect_intrusive_heap.h b/src/dmclock/support/src/indirect_intrusive_heap.h new file mode 100644 index 000000000..d84a48784 --- /dev/null +++ b/src/dmclock/support/src/indirect_intrusive_heap.h @@ -0,0 +1,567 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <memory> +#include <vector> +#include <string> +#include <iostream> +#include <functional> +#include <algorithm> + +#include "assert.h" + + +namespace crimson { + using IndIntruHeapData = size_t; + + /* T is the ultimate data that's being stored in the heap, although + * through indirection. + * + * I is the indirect type that will actually be stored in the heap + * and that must allow dereferencing (via operator*) to yield a + * T&. + * + * C is a functor when given two T&'s will return true if the first + * must precede the second. + * + * heap_info is a data member pointer as to where the heap data in T + * is stored. + * + * K is the branching factor of the heap, default is 2 (binary heap). + */ + template<typename I, + typename T, + IndIntruHeapData T::*heap_info, + typename C, + unsigned K = 2> + class IndIntruHeap { + + // shorthand + using HeapIndex = IndIntruHeapData; + + static_assert( + std::is_same<T,typename std::pointer_traits<I>::element_type>::value, + "class I must resolve to class T by indirection (pointer dereference)"); + + static_assert( + std::is_same<bool, + typename std::result_of<C(const T&,const T&)>::type>::value, + "class C must define operator() to take two const T& and return a bool"); + + static_assert(K >= 2, "K (degree of branching) must be at least 2"); + + class Iterator { + friend IndIntruHeap<I, T, heap_info, C, K>; + + IndIntruHeap<I, T, heap_info, C, K>* heap; + HeapIndex index; + + Iterator(IndIntruHeap<I, T, heap_info, C, K>& _heap, HeapIndex _index) : + heap(&_heap), + index(_index) + { + // empty + } + + public: + + Iterator(Iterator&& other) : + heap(other.heap), + index(other.index) + { + // empty + } + + Iterator(const Iterator& other) : + heap(other.heap), + index(other.index) + { + // empty + } + + Iterator& operator=(Iterator&& other) { + std::swap(heap, other.heap); + std::swap(index, other.index); + return *this; + } + + Iterator& operator=(const Iterator& other) { + heap = other.heap; + index = other.index; + } + + Iterator& operator++() { + if (index <= heap->count) { + ++index; + } + return *this; + } + + bool operator==(const Iterator& other) const { + return heap == other.heap && index == other.index; + } + + bool operator!=(const Iterator& other) const { + return !(*this == other); + } + + T& operator*() { + return *heap->data[index]; + } + + T* operator->() { + return &(*heap->data[index]); + } + +#if 0 + // the item this iterator refers to + void increase() { + heap.sift_up(index); + } +#endif + }; // class Iterator + + + class ConstIterator { + friend IndIntruHeap<I, T, heap_info, C, K>; + + const IndIntruHeap<I, T, heap_info, C, K>* heap; + HeapIndex index; + + ConstIterator(const IndIntruHeap<I, T, heap_info, C, K>& _heap, + HeapIndex _index) : + heap(&_heap), + index(_index) + { + // empty + } + + public: + + ConstIterator(ConstIterator&& other) : + heap(other.heap), + index(other.index) + { + // empty + } + + ConstIterator(const ConstIterator& other) : + heap(other.heap), + index(other.index) + { + // empty + } + + ConstIterator& operator=(ConstIterator&& other) { + std::swap(heap, other.heap); + std::swap(index, other.index); + return *this; + } + + ConstIterator& operator=(const ConstIterator& other) { + heap = other.heap; + index = other.index; + } + + ConstIterator& operator++() { + if (index <= heap->count) { + ++index; + } + return *this; + } + + bool operator==(const ConstIterator& other) const { + return heap == other.heap && index == other.index; + } + + bool operator!=(const ConstIterator& other) const { + return !(*this == other); + } + + const T& operator*() { + return *heap->data[index]; + } + + const T* operator->() { + return &(*heap->data[index]); + } + }; // class ConstIterator + + + protected: + + std::vector<I> data; + HeapIndex count; + C comparator; + + public: + + IndIntruHeap() : + count(0) + { + // empty + } + + IndIntruHeap(const IndIntruHeap<I,T,heap_info,C,K>& other) : + count(other.count) + { + for (HeapIndex i = 0; i < other.count; ++i) { + data.push_back(other.data[i]); + } + } + + bool empty() const { return 0 == count; } + + size_t size() const { return (size_t) count; } + + T& top() { return *data[0]; } + + const T& top() const { return *data[0]; } + + I& top_ind() { return data[0]; } + + const I& top_ind() const { return data[0]; } + + void push(I&& item) { + HeapIndex i = count++; + intru_data_of(item) = i; + data.emplace_back(std::move(item)); + sift_up(i); + } + + void push(const I& item) { + I copy(item); + push(std::move(copy)); + } + + void pop() { + remove(HeapIndex(0)); + } + + void remove(Iterator& i) { + remove(i.index); + i = end(); + } + + Iterator find(const I& ind_item) { + for (HeapIndex i = 0; i < count; ++i) { + if (data[i] == ind_item) { + return Iterator(*this, i); + } + } + return end(); + } + + // when passing in value we do a comparison via operator== + Iterator find(const T& item) { + for (HeapIndex i = 0; i < count; ++i) { + if (*data[i] == item) { + return Iterator(*this, i); + } + } + return end(); + } + + // reverse find -- start looking from bottom of heap + Iterator rfind(const I& ind_item) { + // HeapIndex is unsigned, so we can't allow to go negative; so + // we'll keep it one more than actual index + for (HeapIndex i = count; i > 0; --i) { + if (data[i-1] == ind_item) { + return Iterator(*this, i-1); + } + } + return end(); + } + + // reverse find -- start looking from bottom of heap + Iterator rfind(const T& item) { + // HeapIndex is unsigned, so we can't allow to go negative; so + // we'll keep it one more than actual index + for (HeapIndex i = count; i > 0; --i) { + if (*data[i-1] == item) { + return Iterator(*this, i-1); + } + } + return end(); + } + + ConstIterator find(const I& ind_item) const { + for (HeapIndex i = 0; i < count; ++i) { + if (data[i] == ind_item) { + return ConstIterator(*this, i); + } + } + return cend(); + } + + // when passing in value we do a comparison via operator== + ConstIterator find(const T& item) const { + for (HeapIndex i = 0; i < count; ++i) { + if (*data[i] == item) { + return ConstIterator(*this, i); + } + } + return cend(); + } + + // reverse find -- start looking from bottom of heap + ConstIterator rfind(const I& ind_item) const { + // HeapIndex is unsigned, so we can't allow to go negative; so + // we'll keep it one more than actual index + for (HeapIndex i = count; i > 0; --i) { + if (data[i-1] == ind_item) { + return ConstIterator(*this, i-1); + } + } + return cend(); + } + + // reverse find -- start looking from bottom of heap + ConstIterator rfind(const T& item) const { + // HeapIndex is unsigned, so we can't allow to go negative; so + // we'll keep it one more than actual index + for (HeapIndex i = count; i > 0; --i) { + if (*data[i-1] == item) { + return ConstIterator(*this, i-1); + } + } + return cend(); + } + + Iterator at(const I& ind_item) { + auto ind = intru_data_of(ind_item); + if (ind >= count) { + throw std::out_of_range( + std::to_string(ind) + " >= " + std::to_string(count)); + } + assert(data[ind] == ind_item); + return Iterator(*this, ind); + } + + void promote(T& item) { + sift_up(item.*heap_info); + } + + void demote(T& item) { + sift_down(item.*heap_info); + } + + void adjust(T& item) { + sift(item.*heap_info); + } + + Iterator begin() { + return Iterator(*this, 0); + } + + Iterator end() { + return Iterator(*this, count); + } + + ConstIterator cbegin() const { + return ConstIterator(*this, 0); + } + + ConstIterator cend() const { + return ConstIterator(*this, count); + } + + friend std::ostream& operator<<(std::ostream& out, const IndIntruHeap& h) { + auto i = h.data.cbegin(); + if (i != h.data.cend()) { + out << **i; + ++i; + while (i != h.data.cend()) { + out << ", " << **i; + } + } + return out; + } + + // can only be called if I is copiable; copies heap into a vector + // and sorts it before displaying it + std::ostream& + display_sorted(std::ostream& out, + std::function<bool(const T&)> filter = all_filter) const { + static_assert(std::is_copy_constructible<I>::value, + "cannot call display_sorted when class I is not copy" + " constructible"); + auto compare = [this] (const I first, const I second) -> bool { + return this->comparator(*first, *second); + }; + std::vector<I> copy(data); + std::sort(copy.begin(), copy.end(), compare); + + bool first = true; + for (auto c = copy.begin(); c != copy.end(); ++c) { + if (filter(**c)) { + if (!first) { + out << ", "; + } else { + first = false; + } + out << **c; + } + } + + return out; + } + + + protected: + + static IndIntruHeapData& intru_data_of(const I& item) { + return (*item).*heap_info; + } + + void remove(HeapIndex i) { + std::swap(data[i], data[--count]); + intru_data_of(data[i]) = i; + + // the following needs to be sift (and not sift_down) as it can + // go up or down the heap; imagine the heap vector contains 0, + // 10, 100, 20, 30, 200, 300, 40; then 200 is removed, and 40 + // would have to be sifted upwards + // sift(i); + sift(i); + + data.pop_back(); + } + + // default value of filter parameter to display_sorted + static bool all_filter(const T& data) { return true; } + + // when i is negative? + static inline HeapIndex parent(HeapIndex i) { + assert(0 != i); + return (i - 1) / K; + } + + // index of left child when K==2, index of left-most child when K>2 + static inline HeapIndex lhs(HeapIndex i) { return K*i + 1; } + + // index of right child when K==2, index of right-most child when K>2 + static inline HeapIndex rhs(HeapIndex i) { return K*i + K; } + + void sift_up(HeapIndex i) { + while (i > 0) { + HeapIndex pi = parent(i); + if (!comparator(*data[i], *data[pi])) { + break; + } + + std::swap(data[i], data[pi]); + intru_data_of(data[i]) = i; + intru_data_of(data[pi]) = pi; + i = pi; + } + } // sift_up + + // use this sift_down definition when K>2; it's more general and + // uses a loop; EnableBool insures template uses a template + // parameter + template<bool EnableBool=true> + typename std::enable_if<(K>2)&&EnableBool,void>::type sift_down(HeapIndex i) { + if (i >= count) return; + while (true) { + HeapIndex li = lhs(i); + + if (li < count) { + HeapIndex ri = std::min(rhs(i), count - 1); + + // find the index of min. child + HeapIndex min_i = li; + for (HeapIndex k = li + 1; k <= ri; ++k) { + if (comparator(*data[k], *data[min_i])) { + min_i = k; + } + } + + if (comparator(*data[min_i], *data[i])) { + std::swap(data[i], data[min_i]); + intru_data_of(data[i]) = i; + intru_data_of(data[min_i]) = min_i; + i = min_i; + } else { + // no child is smaller + break; + } + } else { + // no children + break; + } + } + } // sift_down + + // use this sift_down definition when K==2; EnableBool insures + // template uses a template parameter + template<bool EnableBool=true> + typename std::enable_if<K==2&&EnableBool,void>::type sift_down(HeapIndex i) { + if (i >= count) return; + while (true) { + const HeapIndex li = lhs(i); + const HeapIndex ri = 1 + li; + + if (li < count) { + if (comparator(*data[li], *data[i])) { + if (ri < count && comparator(*data[ri], *data[li])) { + std::swap(data[i], data[ri]); + intru_data_of(data[i]) = i; + intru_data_of(data[ri]) = ri; + i = ri; + } else { + std::swap(data[i], data[li]); + intru_data_of(data[i]) = i; + intru_data_of(data[li]) = li; + i = li; + } + } else if (ri < count && comparator(*data[ri], *data[i])) { + std::swap(data[i], data[ri]); + intru_data_of(data[i]) = i; + intru_data_of(data[ri]) = ri; + i = ri; + } else { + // no child is smaller + break; + } + } else { + // no children + break; + } + } // while + } // sift_down + + void sift(HeapIndex i) { + if (i == 0) { + // if we're at top, can only go down + sift_down(i); + } else { + HeapIndex pi = parent(i); + if (comparator(*data[i], *data[pi])) { + // if we can go up, we will + sift_up(i); + } else { + // otherwise we'll try to go down + sift_down(i); + } + } + } // sift + }; // class IndIntruHeap + +} // namespace crimson diff --git a/src/dmclock/support/src/intrusive_heap.h b/src/dmclock/support/src/intrusive_heap.h new file mode 100644 index 000000000..21d3ea9a0 --- /dev/null +++ b/src/dmclock/support/src/intrusive_heap.h @@ -0,0 +1,221 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <vector> +#include <string> +#include <iostream> +#include <functional> + +#include "assert.h" + + +namespace crimson { + using IntruHeapData = size_t; + + // T = type of data in heap; I = functor that returns a non-const + // reference to IntruHeapData; C = functor that compares two const + // refs and return true if the first precedes the second + template<typename T, typename I, typename C> + class IntruHeap { + + static_assert( + std::is_same<IntruHeapData&,typename std::result_of<I(T&)>::type>::value, + "class I must define operator() to take T& and return a IntruHeapData&."); + + static_assert( + std::is_same<bool,typename std::result_of<C(const T&,const T&)>::type>::value, + "class C must define operator() to take two const T& and return a bool."); + + + protected: + using index_t = IntruHeapData; + + std::vector<T> data; + index_t count; + I intru_data_of; + C comparator; + + public: + + IntruHeap() : + count(0) + { + // empty + } + + IntruHeap(const IntruHeap<T,I,C>& other) : + count(other.count) + { + for (uint i = 0; i < other.count; ++i) { + data.push_back(other.data[i]); + } + } + + bool empty() const { return 0 == count; } + + T& top() { return data[0]; } + + void push(T&& item) { + index_t i = count++; + intru_data_of(item) = i; + data.emplace_back(item); + sift_up(i); + } + + void push(const T& item) { + T copy(item); + push(std::move(copy)); + } + + void pop() { + std::swap(data[0], data[--count]); + intru_data_of(data[0]) = 0; + data.pop_back(); + sift_down(0); + } + + void adjust_up(T& item) { + sift_up(intru_data_of(item)); + } + + void adjust_down(T& item) { + sift_down(intru_data_of(item)); + } + + void adjust(T& item) { + sift(intru_data_of(item)); + } + + friend std::ostream& operator<<(std::ostream& out, const IntruHeap& h) { + for (uint i = 0; i < h.count; ++i) { + out << h.data[i] << ", "; + } + return out; + } + + std::ostream& + display_sorted(std::ostream& out, + bool insert_line_breaks = true, + std::function<bool(const T&)> filter = all_filter) const { + IntruHeap<T,I,C> copy = *this; + + bool first = true; + out << "[ "; + + while(!copy.empty()) { + const T& top = copy.top(); + if (filter(top)) { + if (!first) { + out << ", "; + } + if (insert_line_breaks) { + out << std::endl << " "; + } + out << copy.top(); + first = false; + } + copy.pop(); + } + + out << " ]"; + if (insert_line_breaks) { + out << std::endl; + } + + return out; + } + + + protected: + + // default value of filter parameter to display_sorted + static bool all_filter(const T& data) { return true; } + + // when i is negative? + static inline index_t parent(index_t i) { + assert(0 != i); + return (i - 1) / 2; + } + + static inline index_t lhs(index_t i) { return 2*i + 1; } + + static inline index_t rhs(index_t i) { return 2*i + 2; } + + void sift_up(index_t i) { + while (i > 0) { + index_t pi = parent(i); + if (!comparator(data[i], data[pi])) { + break; + } + + std::swap(data[i], data[pi]); + intru_data_of(data[i]) = i; + intru_data_of(data[pi]) = pi; + i = pi; + } + } // sift_up + + void sift_down(index_t i) { + while (i < count) { + index_t li = lhs(i); + index_t ri = rhs(i); + + if (li < count) { + if (comparator(data[li], data[i])) { + if (ri < count && comparator(data[ri], data[li])) { + std::swap(data[i], data[ri]); + intru_data_of(data[i]) = i; + intru_data_of(data[ri]) = ri; + i = ri; + } else { + std::swap(data[i], data[li]); + intru_data_of(data[i]) = i; + intru_data_of(data[li]) = li; + i = li; + } + } else if (ri < count && comparator(data[ri], data[i])) { + std::swap(data[i], data[ri]); + intru_data_of(data[i]) = i; + intru_data_of(data[ri]) = ri; + i = ri; + } else { + break; + } + } else { + break; + } + } + } // sift_down + + void sift(index_t i) { + if (i == 0) { + // if we're at top, can only go down + sift_down(i); + } else { + index_t pi = parent(i); + if (comparator(data[i], data[pi])) { + // if we can go up, we will + sift_up(i); + } else { + // otherwise we'll try to go down + sift_down(i); + } + } + } // sift + }; // class IntruHeap +} // namespace crimson diff --git a/src/dmclock/support/src/profile.h b/src/dmclock/support/src/profile.h new file mode 100644 index 000000000..8b357dbfc --- /dev/null +++ b/src/dmclock/support/src/profile.h @@ -0,0 +1,121 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + + +#include <cmath> +#include <chrono> + + +namespace crimson { + template<typename T> + class ProfileBase { + + protected: + + using clock = std::chrono::steady_clock; + + uint count = 0; + typename T::rep sum = 0; + typename T::rep sum_squares = 0; + typename T::rep low = 0; + typename T::rep high = 0; + + public: + + uint get_count() const { return count; } + typename T::rep get_sum() const { return sum; } + typename T::rep get_low() const { return low; } + typename T::rep get_high() const { return high; } + double get_mean() const { + if (0 == count) return nan(""); + return sum / double(count); } + double get_std_dev() const { + if (0 == count) return nan(""); + double variance = + (count * sum_squares - sum * sum) / double(count * count); + return sqrt(variance); + } + }; // class ProfileBase + + + // forward declaration for friend + template<typename T> + class ProfileCombiner; + + + template<typename T> + class ProfileTimer : public ProfileBase<T> { + friend ProfileCombiner<T>; + + using super = ProfileBase<T>; + + bool is_timing = false; + typename super::clock::time_point start_time; + + public: + + ProfileTimer() { + } + + void start() { + assert(!is_timing); + start_time = super::clock::now(); + is_timing = true; + } + + void stop() { + assert(is_timing); + T duration = std::chrono::duration_cast<T>(super::clock::now() - start_time); + typename T::rep duration_count = duration.count(); + this->sum += duration_count; + this->sum_squares += duration_count * duration_count; + if (0 == this->count) { + this->low = duration_count; + this->high = duration_count; + } else { + if (duration_count < this->low) this->low = duration_count; + else if (duration_count > this->high) this->high = duration_count; + } + ++this->count; + is_timing = false; + } + }; // class ProfileTimer + + + template<typename T> + class ProfileCombiner : public ProfileBase<T> { + + using super = ProfileBase<T>; + + public: + + ProfileCombiner() {} + + void combine(const ProfileTimer<T>& timer) { + if (0 == this->count) { + this->low = timer.low; + this->high = timer.high; + } else { + if (timer.low < this->low) this->low = timer.low; + else if (timer.high > this->high) this->high = timer.high; + } + this->count += timer.count; + this->sum += timer.sum; + this->sum_squares += timer.sum_squares; + } + }; // class ProfileCombiner +} // namespace crimson diff --git a/src/dmclock/support/src/run_every.cc b/src/dmclock/support/src/run_every.cc new file mode 100644 index 000000000..14f1452b9 --- /dev/null +++ b/src/dmclock/support/src/run_every.cc @@ -0,0 +1,94 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2017 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include "run_every.h" + + +// can define ADD_MOVE_SEMANTICS, although not fully debugged and tested + + +namespace chrono = std::chrono; + + +#ifdef ADD_MOVE_SEMANTICS +crimson::RunEvery::RunEvery() +{ + // empty +} + + +crimson::RunEvery& crimson::RunEvery::operator=(crimson::RunEvery&& other) +{ + // finish run every thread + { + Guard g(mtx); + finishing = true; + cv.notify_one(); + } + if (thd.joinable()) { + thd.join(); + } + + // transfer info over from previous thread + finishing.store(other.finishing); + wait_period = other.wait_period; + body = other.body; + + // finish other thread + other.finishing.store(true); + other.cv.notify_one(); + + // start this thread + thd = std::thread(&RunEvery::run, this); + + return *this; +} +#endif + + +crimson::RunEvery::~RunEvery() { + join(); +} + + +void crimson::RunEvery::join() { + { + Guard l(mtx); + if (finishing) return; + finishing = true; + cv.notify_all(); + } + thd.join(); +} + +// mtx must be held by caller +void crimson::RunEvery::try_update(milliseconds _wait_period) { + if (_wait_period != wait_period) { + wait_period = _wait_period; + } +} + +void crimson::RunEvery::run() { + Lock l(mtx); + while(!finishing) { + TimePoint until = chrono::steady_clock::now() + wait_period; + while (!finishing && chrono::steady_clock::now() < until) { + cv.wait_until(l, until); + } + if (!finishing) { + body(); + } + } +} diff --git a/src/dmclock/support/src/run_every.h b/src/dmclock/support/src/run_every.h new file mode 100644 index 000000000..f93961db1 --- /dev/null +++ b/src/dmclock/support/src/run_every.h @@ -0,0 +1,81 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2017 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#pragma once + +#include <chrono> +#include <mutex> +#include <condition_variable> +#include <thread> +#include <functional> + + +namespace crimson { + using std::chrono::duration_cast; + using std::chrono::milliseconds; + + // runs a given simple function object waiting wait_period + // milliseconds between; the destructor stops the other thread + // immediately + class RunEvery { + using Lock = std::unique_lock<std::mutex>; + using Guard = std::lock_guard<std::mutex>; + using TimePoint = std::chrono::steady_clock::time_point; + + bool finishing = false; + std::chrono::milliseconds wait_period; + std::function<void()> body; + std::mutex mtx; + std::condition_variable cv; + + // put threads last so all other variables are initialized first + + std::thread thd; + + public: + +#ifdef ADD_MOVE_SEMANTICS + RunEvery(); +#endif + + template<typename D> + RunEvery(D _wait_period, + const std::function<void()>& _body) : + wait_period(duration_cast<milliseconds>(_wait_period)), + body(_body) + { + thd = std::thread(&RunEvery::run, this); + } + + RunEvery(const RunEvery& other) = delete; + RunEvery& operator=(const RunEvery& other) = delete; + RunEvery(RunEvery&& other) = delete; +#ifdef ADD_MOVE_SEMANTICS + RunEvery& operator=(RunEvery&& other); +#else + RunEvery& operator=(RunEvery&& other) = delete; +#endif + + ~RunEvery(); + + void join(); + // update wait period in milliseconds + void try_update(milliseconds _wait_period); + + protected: + + void run(); + }; +} diff --git a/src/dmclock/support/test/CMakeLists.txt b/src/dmclock/support/test/CMakeLists.txt new file mode 100644 index 000000000..24b352157 --- /dev/null +++ b/src/dmclock/support/test/CMakeLists.txt @@ -0,0 +1,26 @@ +include_directories(../src) + +set(local_flags "-Wall -pthread") + +# dmclock does not use intrusive heap (but it does use indirect +# intrusive heap), so we won't use this code +if(false) + set(srcs + test_intrusive_heap.cc) + add_executable(test_intru_heap test_intrusive_heap.cc) + set_source_files_properties(${srcs} + PROPERTIES + COMPILE_FLAGS "${local_flags}") +endif(false) + +set(test_srcs test_indirect_intrusive_heap.cc) + +set_source_files_properties(${test_srcs} + PROPERTIES + COMPILE_FLAGS "${local_flags}" + ) + +add_executable(dmclock-data-struct-tests ${test_srcs}) + +target_link_libraries(dmclock-data-struct-tests + LINK_PRIVATE GTest::GTest GTest::Main Threads::Threads) diff --git a/src/dmclock/support/test/test_ind_intru_heap.cc b/src/dmclock/support/test/test_ind_intru_heap.cc new file mode 100644 index 000000000..8e7ee6931 --- /dev/null +++ b/src/dmclock/support/test/test_ind_intru_heap.cc @@ -0,0 +1,89 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include <memory> +#include <string> +#include <iostream> + +#include "indirect_intrusive_heap.h" + + +class TestCompare; + + +class Test1 { + friend TestCompare; + + int data; + +public: + + crimson::IndIntruHeapData heap_data; + + explicit Test1(int _data) : data(_data) {} + + friend std::ostream& operator<<(std::ostream& out, const Test1& d) { + out << d.data << " (" << d.heap_data << ")"; + return out; + } + + int& the_data() { return data; } +}; + + +struct TestCompare { + bool operator()(const Test1& d1, const Test1& d2) { + return d1.data < d2.data; + } +}; + + +int main(int argc, char** argv) { + Test1 d1(2); + Test1 d2(3); + Test1 d3(1); + Test1 d4(-5); + + crimson::IndIntruHeap<std::shared_ptr<Test1>, Test1, &Test1::heap_data, TestCompare> my_heap; + + const std::shared_ptr<Test1> d99 = std::make_shared<Test1>(99); + + my_heap.push(std::make_shared<Test1>(2)); + my_heap.push(d99); + my_heap.push(std::make_shared<Test1>(1)); + my_heap.push(std::make_shared<Test1>(-5)); + my_heap.push(std::make_shared<Test1>(12)); + my_heap.push(std::make_shared<Test1>(-12)); + my_heap.push(std::make_shared<Test1>(-7)); + + std::cout << my_heap << std::endl; + + auto& t = my_heap.top(); + t.the_data() = 17; + my_heap.adjust_down(t); + + std::cout << my_heap << std::endl; + + my_heap.display_sorted(std::cout); + + while (!my_heap.empty()) { + auto& top = my_heap.top(); + std::cout << top << std::endl; + my_heap.pop(); + std::cout << my_heap << std::endl; + } + + return 0; +} diff --git a/src/dmclock/support/test/test_indirect_intrusive_heap.cc b/src/dmclock/support/test/test_indirect_intrusive_heap.cc new file mode 100644 index 000000000..e74c3181e --- /dev/null +++ b/src/dmclock/support/test/test_indirect_intrusive_heap.cc @@ -0,0 +1,1022 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include <iostream> +#include <memory> +#include <set> +#include <algorithm> +#include <random> + +#include "gtest/gtest.h" + +#include "indirect_intrusive_heap.h" + + +struct Elem { + int data; + + crimson::IndIntruHeapData heap_data; + crimson::IndIntruHeapData heap_data_alt; + + explicit Elem(int _data) : data(_data) { } + + bool operator==(const Elem& other) const { + return data == other.data; + } + + bool operator<(const Elem& other) const { + return data < other.data; + } + + friend std::ostream& operator<<(std::ostream& out, const Elem& d) { + out << d.data; + return out; + } +}; + + +// sorted low to high +struct ElemCompare { + bool operator()(const Elem& d1, const Elem& d2) const { + return d1.data < d2.data; + } +}; + + +// first all evens precede all odds, then they're sorted high to low +struct ElemCompareAlt { + bool operator()(const Elem& d1, const Elem& d2) { + if (0 == d1.data % 2) { + if (0 == d2.data % 2) { + return d1.data > d2.data; + } else { + return true; + } + } else if (0 == d2.data % 2) { + return false; + } else { + return d1.data > d2.data; + } + } +}; + + +class HeapFixture1: public ::testing::Test { + +public: + + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare> heap; + + std::shared_ptr<Elem> data1, data2, data3, data4, data5, data6, data7; + + void SetUp() { + data1 = std::make_shared<Elem>(2); + data2 = std::make_shared<Elem>(99); + data3 = std::make_shared<Elem>(1); + data4 = std::make_shared<Elem>(-5); + data5 = std::make_shared<Elem>(12); + data6 = std::make_shared<Elem>(-12); + data7 = std::make_shared<Elem>(-7); + + heap.push(data1); + heap.push(data2); + heap.push(data3); + heap.push(data4); + heap.push(data5); + heap.push(data6); + heap.push(data7); + } + + void TearDown() { + // nothing to do + } +}; // class HeapFixture1 + +TEST(IndIntruHeap, shared_ptr) { + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare> heap; + + EXPECT_TRUE(heap.empty()); + + heap.push(std::make_shared<Elem>(2)); + + EXPECT_FALSE(heap.empty()); + + heap.push(std::make_shared<Elem>(99)); + heap.push(std::make_shared<Elem>(1)); + heap.push(std::make_shared<Elem>(-5)); + heap.push(std::make_shared<Elem>(12)); + heap.push(std::make_shared<Elem>(-12)); + heap.push(std::make_shared<Elem>(-7)); + + // std::cout << heap << std::endl; + + EXPECT_FALSE(heap.empty()); + + EXPECT_EQ(-12, heap.top().data); + heap.pop(); + EXPECT_EQ(-7, heap.top().data); + heap.pop(); + EXPECT_EQ(-5, heap.top().data); + heap.pop(); + EXPECT_EQ(1, heap.top().data); + heap.pop(); + EXPECT_EQ(2, heap.top().data); + heap.pop(); + EXPECT_EQ(12, heap.top().data); + heap.pop(); + EXPECT_EQ(99, heap.top().data); + + EXPECT_FALSE(heap.empty()); + heap.pop(); + EXPECT_TRUE(heap.empty()); +} + + +TEST(IndIntruHeap, unique_ptr) { + crimson::IndIntruHeap<std::unique_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare> heap; + + EXPECT_TRUE(heap.empty()); + + heap.push(std::unique_ptr<Elem>(new Elem(2))); + + EXPECT_FALSE(heap.empty()); + + heap.push(std::unique_ptr<Elem>(new Elem(99))); + heap.push(std::unique_ptr<Elem>(new Elem(1))); + heap.push(std::unique_ptr<Elem>(new Elem(-5))); + heap.push(std::unique_ptr<Elem>(new Elem(12))); + heap.push(std::unique_ptr<Elem>(new Elem(-12))); + heap.push(std::unique_ptr<Elem>(new Elem(-7))); + + EXPECT_FALSE(heap.empty()); + + EXPECT_EQ(-12, heap.top().data); + heap.pop(); + EXPECT_EQ(-7, heap.top().data); + heap.pop(); + EXPECT_EQ(-5, heap.top().data); + heap.pop(); + EXPECT_EQ(1, heap.top().data); + heap.pop(); + EXPECT_EQ(2, heap.top().data); + heap.pop(); + EXPECT_EQ(12, heap.top().data); + heap.pop(); + EXPECT_EQ(99, heap.top().data); + + EXPECT_FALSE(heap.empty()); + heap.pop(); + EXPECT_TRUE(heap.empty()); +} + + +TEST(IndIntruHeap, regular_ptr) { + crimson::IndIntruHeap<Elem*, Elem, &Elem::heap_data, ElemCompare> heap; + + EXPECT_TRUE(heap.empty()); + + heap.push(new Elem(2)); + + EXPECT_FALSE(heap.empty()); + + heap.push(new Elem(99)); + heap.push(new Elem(1)); + heap.push(new Elem(-5)); + heap.push(new Elem(12)); + heap.push(new Elem(-12)); + heap.push(new Elem(-7)); + + EXPECT_FALSE(heap.empty()); + + EXPECT_EQ(-12, heap.top().data); + { + auto i = &heap.top(); + heap.pop(); + delete i; + } + + EXPECT_EQ(-7, heap.top().data); + { + auto i = &heap.top(); + heap.pop(); + delete i; + } + + EXPECT_EQ(-5, heap.top().data); + { + auto i = &heap.top(); + heap.pop(); + delete i; + } + + EXPECT_EQ(1, heap.top().data); + { + auto i = &heap.top(); + heap.pop(); + delete i; + } + + EXPECT_EQ(2, heap.top().data); + { + auto i = &heap.top(); + heap.pop(); + delete i; + } + + EXPECT_EQ(12, heap.top().data); + { + auto i = &heap.top(); + heap.pop(); + delete i; + } + + EXPECT_EQ(99, heap.top().data); + { + auto i = &heap.top(); + EXPECT_FALSE(heap.empty()); + heap.pop(); + delete i; + } + + EXPECT_TRUE(heap.empty()); +} + + +TEST(IndIntruHeap, K_3) { + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 3> heap; + + EXPECT_TRUE(heap.empty()); + + heap.push(std::make_shared<Elem>(2)); + + EXPECT_FALSE(heap.empty()); + + heap.push(std::make_shared<Elem>(99)); + heap.push(std::make_shared<Elem>(1)); + heap.push(std::make_shared<Elem>(-5)); + heap.push(std::make_shared<Elem>(12)); + heap.push(std::make_shared<Elem>(-12)); + heap.push(std::make_shared<Elem>(-7)); + + // std::cout << heap << std::endl; + + EXPECT_FALSE(heap.empty()); + + EXPECT_EQ(-12, heap.top().data); + heap.pop(); + EXPECT_EQ(-7, heap.top().data); + heap.pop(); + EXPECT_EQ(-5, heap.top().data); + heap.pop(); + EXPECT_EQ(1, heap.top().data); + heap.pop(); + EXPECT_EQ(2, heap.top().data); + heap.pop(); + EXPECT_EQ(12, heap.top().data); + heap.pop(); + EXPECT_EQ(99, heap.top().data); + + EXPECT_FALSE(heap.empty()); + heap.pop(); + EXPECT_TRUE(heap.empty()); +} + + +TEST(IndIntruHeap, K_4) { + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 4> heap; + + EXPECT_TRUE(heap.empty()); + + heap.push(std::make_shared<Elem>(2)); + + EXPECT_FALSE(heap.empty()); + + heap.push(std::make_shared<Elem>(99)); + heap.push(std::make_shared<Elem>(1)); + heap.push(std::make_shared<Elem>(-5)); + heap.push(std::make_shared<Elem>(12)); + heap.push(std::make_shared<Elem>(-12)); + heap.push(std::make_shared<Elem>(-7)); + + // std::cout << heap << std::endl; + + EXPECT_FALSE(heap.empty()); + + EXPECT_EQ(-12, heap.top().data); + heap.pop(); + EXPECT_EQ(-7, heap.top().data); + heap.pop(); + EXPECT_EQ(-5, heap.top().data); + heap.pop(); + EXPECT_EQ(1, heap.top().data); + heap.pop(); + EXPECT_EQ(2, heap.top().data); + heap.pop(); + EXPECT_EQ(12, heap.top().data); + heap.pop(); + EXPECT_EQ(99, heap.top().data); + + EXPECT_FALSE(heap.empty()); + heap.pop(); + EXPECT_TRUE(heap.empty()); +} + + +TEST(IndIntruHeap, K_10) { + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 10> heap; + + EXPECT_TRUE(heap.empty()); + + heap.push(std::make_shared<Elem>(2)); + + EXPECT_FALSE(heap.empty()); + + heap.push(std::make_shared<Elem>(99)); + heap.push(std::make_shared<Elem>(1)); + heap.push(std::make_shared<Elem>(-5)); + heap.push(std::make_shared<Elem>(12)); + heap.push(std::make_shared<Elem>(-12)); + heap.push(std::make_shared<Elem>(-7)); + + // std::cout << heap << std::endl; + + EXPECT_FALSE(heap.empty()); + + EXPECT_EQ(-12, heap.top().data); + heap.pop(); + EXPECT_EQ(-7, heap.top().data); + heap.pop(); + EXPECT_EQ(-5, heap.top().data); + heap.pop(); + EXPECT_EQ(1, heap.top().data); + heap.pop(); + EXPECT_EQ(2, heap.top().data); + heap.pop(); + EXPECT_EQ(12, heap.top().data); + heap.pop(); + EXPECT_EQ(99, heap.top().data); + + EXPECT_FALSE(heap.empty()); + heap.pop(); + EXPECT_TRUE(heap.empty()); +} + + +TEST(IndIntruHeap, multi_K) { + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 2> heap2; + + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 3> heap3; + + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 4> heap4; + + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 10> heap10; + + // 250 should give us at least 4 levels on all heaps + constexpr size_t count = 250; + + std::srand(std::time(0)); // use current time as seed for random generator + + // insert same set of random values into the four heaps + for (size_t i = 0; i < count; ++i) { + int value = std::rand() % 201 - 100; // -100...+100 + auto data = std::make_shared<Elem>(value); + heap2.push(data); + heap3.push(data); + heap4.push(data); + heap10.push(data); + } + + auto bound = std::numeric_limits<decltype(Elem::data)>::min(); + + for (size_t i = 0; i < count; ++i) { + auto current = heap2.top().data; + + EXPECT_GE(current, bound) << + "we should never go down, only increase or remain the same"; + EXPECT_EQ(current, heap3.top().data) << + "heap1's data and heap3's data should match"; + EXPECT_EQ(current, heap4.top().data) << + "heap1's data and heap4's data should match"; + EXPECT_EQ(current, heap10.top().data) << + "heap1's data and heap10's data should match"; + + heap2.pop(); + heap3.pop(); + heap4.pop(); + heap10.pop(); + + bound = current; + } + + EXPECT_TRUE(heap2.empty()) << "should be empty after all elements popped"; + EXPECT_TRUE(heap3.empty()) << "should be empty after all elements popped"; + EXPECT_TRUE(heap4.empty()) << "should be empty after all elements popped"; + EXPECT_TRUE(heap10.empty()) << "should be empty after all elements popped"; +} + + +TEST(IndIntruHeap, demote) { + crimson::IndIntruHeap<std::unique_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare> heap; + + heap.push(std::unique_ptr<Elem>(new Elem(2))); + heap.push(std::unique_ptr<Elem>(new Elem(99))); + heap.push(std::unique_ptr<Elem>(new Elem(1))); + heap.push(std::unique_ptr<Elem>(new Elem(-5))); + heap.push(std::unique_ptr<Elem>(new Elem(12))); + heap.push(std::unique_ptr<Elem>(new Elem(-12))); + heap.push(std::unique_ptr<Elem>(new Elem(-7))); + + heap.top().data = 24; + + heap.demote(heap.top()); + + EXPECT_EQ(-7, heap.top().data); + + heap.pop(); + heap.pop(); + heap.pop(); + heap.pop(); + heap.pop(); + + EXPECT_EQ(24, heap.top().data); +} + + +TEST(IndIntruHeap, demote_not) { + crimson::IndIntruHeap<std::unique_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare> heap; + + heap.push(std::unique_ptr<Elem>(new Elem(2))); + heap.push(std::unique_ptr<Elem>(new Elem(99))); + heap.push(std::unique_ptr<Elem>(new Elem(1))); + heap.push(std::unique_ptr<Elem>(new Elem(-5))); + heap.push(std::unique_ptr<Elem>(new Elem(12))); + heap.push(std::unique_ptr<Elem>(new Elem(-12))); + heap.push(std::unique_ptr<Elem>(new Elem(-7))); + + heap.top().data = -99; + + heap.demote(heap.top()); + + EXPECT_EQ(-99, heap.top().data); + + heap.pop(); + + EXPECT_EQ(-7, heap.top().data); +} + + +TEST(IndIntruHeap, promote_and_demote) { + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare> heap; + + auto data1 = std::make_shared<Elem>(1); + + heap.push(std::make_shared<Elem>(2)); + heap.push(std::make_shared<Elem>(99)); + heap.push(data1); + heap.push(std::make_shared<Elem>(-5)); + heap.push(std::make_shared<Elem>(12)); + heap.push(std::make_shared<Elem>(-12)); + heap.push(std::make_shared<Elem>(-7)); + + EXPECT_EQ(-12, heap.top().data); + + data1->data = -99; + heap.promote(*data1); + + EXPECT_EQ(-99, heap.top().data); + + data1->data = 999; + heap.demote(*data1); + + EXPECT_EQ(-12, heap.top().data); + + data1->data = 9; + heap.promote(*data1); + + heap.pop(); // remove -12 + heap.pop(); // remove -7 + heap.pop(); // remove -5 + heap.pop(); // remove 2 + + EXPECT_EQ(9, heap.top().data); +} + + +TEST(IndIntruHeap, adjust) { + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare> heap; + + auto data1 = std::make_shared<Elem>(1); + + heap.push(std::make_shared<Elem>(2)); + heap.push(std::make_shared<Elem>(99)); + heap.push(data1); + heap.push(std::make_shared<Elem>(-5)); + heap.push(std::make_shared<Elem>(12)); + heap.push(std::make_shared<Elem>(-12)); + heap.push(std::make_shared<Elem>(-7)); + + // heap.display_sorted(std::cout); + + EXPECT_EQ(-12, heap.top().data); + + data1->data = 999; + heap.adjust(*data1); + + EXPECT_EQ(-12, heap.top().data); + + data1->data = -99; + heap.adjust(*data1); + + EXPECT_EQ(-99, heap.top().data); + + data1->data = 9; + heap.adjust(*data1); + + EXPECT_EQ(-12, heap.top().data); + + heap.pop(); // remove -12 + heap.pop(); // remove -7 + heap.pop(); // remove -5 + heap.pop(); // remove 2 + + EXPECT_EQ(9, heap.top().data); +} + + +TEST(IndIntruHeap, remove_careful) { + // here we test whether a common mistake in implementing remove is + // done; if after we remove an item and move the last element of the + // heap to the position of the removed element, we need to sift it + // rather than sift_down it. + + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 2> heap; + + heap.push(std::make_shared<Elem>(0)); + heap.push(std::make_shared<Elem>(10)); + heap.push(std::make_shared<Elem>(100)); + heap.push(std::make_shared<Elem>(20)); + heap.push(std::make_shared<Elem>(30)); + heap.push(std::make_shared<Elem>(200)); + heap.push(std::make_shared<Elem>(300)); + heap.push(std::make_shared<Elem>(40)); + + auto k = heap.find(Elem(200)); + EXPECT_NE(heap.end(), k) << + "we should have found an element with the value 200, which we'll remove"; + heap.remove(k); + + auto i = heap.cbegin(); + EXPECT_EQ(0, i->data); + ++i; + EXPECT_EQ(10, i->data); + ++i; + EXPECT_EQ(40, i->data) << + "this needs to be 40 or there's a mistake in implementation"; + ++i; + EXPECT_EQ(20, i->data); + ++i; + EXPECT_EQ(30, i->data); + ++i; + EXPECT_EQ(100, i->data) << + "this needs to be 100 or there's a mistake in implementation"; +} + + +TEST(IndIntruHeap, remove_greatest) { + // See bug #43376 -- removing the greatest element causes an oob + // vector reference + + crimson::IndIntruHeap<std::shared_ptr<Elem>, + Elem, + &Elem::heap_data, + ElemCompare, + 2> heap; + + const int num = 4096; + std::vector<int> toinsert; + toinsert.reserve(num); + std::vector<int> toremove; + toremove.reserve(num - (num/4)); + std::vector<int> tocheck; + tocheck.reserve(num/4); + for (int i = 0; i < num; ++i) { + toinsert.push_back(i); + if (i < (num/2)) { + tocheck.push_back(i); + } else { + toremove.push_back(i); + } + } + + std::default_random_engine generator(0); + std::shuffle( + toinsert.begin(), + toinsert.end(), + generator); + + for (auto i: toinsert) { + heap.push(std::make_shared<Elem>(i)); + } + + for (auto i: toremove) { + auto k = heap.find(Elem(i)); + EXPECT_NE(heap.end(), k) << + "we should have found an element with the value 300, which we'll remove"; + heap.remove(k); + } + + for (auto i: tocheck) { + EXPECT_FALSE(heap.empty()); + EXPECT_EQ(Elem(i), heap.top()); + heap.pop(); + } + EXPECT_TRUE(heap.empty()); +} + + +TEST_F(HeapFixture1, shared_data) { + + crimson::IndIntruHeap<std::shared_ptr<Elem>,Elem,&Elem::heap_data_alt,ElemCompareAlt> heap2; + + heap2.push(data1); + heap2.push(data2); + heap2.push(data3); + heap2.push(data4); + heap2.push(data5); + heap2.push(data6); + heap2.push(data7); + + data3->data = 32; + heap.adjust(*data3); + heap2.adjust(*data3); + + EXPECT_EQ(-12, heap.top().data); + heap.pop(); + EXPECT_EQ(-7, heap.top().data); + heap.pop(); + EXPECT_EQ(-5, heap.top().data); + heap.pop(); + EXPECT_EQ(2, heap.top().data); + heap.pop(); + EXPECT_EQ(12, heap.top().data); + heap.pop(); + EXPECT_EQ(32, heap.top().data); + heap.pop(); + EXPECT_EQ(99, heap.top().data); + + EXPECT_EQ(32, heap2.top().data); + heap2.pop(); + EXPECT_EQ(12, heap2.top().data); + heap2.pop(); + EXPECT_EQ(2, heap2.top().data); + heap2.pop(); + EXPECT_EQ(-12, heap2.top().data); + heap2.pop(); + EXPECT_EQ(99, heap2.top().data); + heap2.pop(); + EXPECT_EQ(-5, heap2.top().data); + heap2.pop(); + EXPECT_EQ(-7, heap2.top().data); +} + + +TEST_F(HeapFixture1, iterator_basics) { + { + unsigned count = 0; + for(auto i = heap.begin(); i != heap.end(); ++i) { + ++count; + } + + EXPECT_EQ(7u, count) << "count should be 7"; + } + + auto i1 = heap.begin(); + + EXPECT_EQ(-12, i1->data) << + "first member with * operator must be smallest"; + + EXPECT_EQ(-12, (*i1).data) << + "first member with -> operator must be smallest"; + + Elem& e1 = *i1; + EXPECT_EQ(-12, e1.data) << + "first member with -> operator must be smallest"; + + { + std::set<int> values; + values.insert(2); + values.insert(99); + values.insert(1); + values.insert(-5); + values.insert(12); + values.insert(-12); + values.insert(-7); + + for(auto i = heap.begin(); i != heap.end(); ++i) { + auto v = *i; + EXPECT_NE(values.end(), values.find(v.data)) << + "value in heap must be part of original set"; + values.erase(v.data); + } + EXPECT_EQ(0u, values.size()) << "all values must have been seen"; + } +} + + +TEST_F(HeapFixture1, const_iterator_basics) { + const auto& cheap = heap; + + { + unsigned count = 0; + for(auto i = cheap.cbegin(); i != cheap.cend(); ++i) { + ++count; + } + + EXPECT_EQ(7u, count) << "count should be 7"; + } + + auto i1 = heap.cbegin(); + + EXPECT_EQ(-12, i1->data) << + "first member with * operator must be smallest"; + + EXPECT_EQ(-12, (*i1).data) << + "first member with -> operator must be smallest"; + + const Elem& e1 = *i1; + EXPECT_EQ(-12, e1.data) << + "first member with -> operator must be smallest"; + + { + std::set<int> values; + values.insert(2); + values.insert(99); + values.insert(1); + values.insert(-5); + values.insert(12); + values.insert(-12); + values.insert(-7); + + for(auto i = heap.cbegin(); i != heap.cend(); ++i) { + auto v = *i; + EXPECT_NE(values.end(), values.find(v.data)) << + "value in heap must be part of original set"; + values.erase(v.data); + } + EXPECT_EQ(0u, values.size()) << "all values must have been seen"; + } +} + + +TEST_F(HeapFixture1, iterator_find_rfind) { + { + auto it1 = heap.find(data7); + EXPECT_NE(heap.end(), it1) << + "find by indirection for included element should succeed"; + EXPECT_EQ(-7, it1->data) << + "find by indirection for included element should result in right value"; + + auto fake_data = std::make_shared<Elem>(-7); + auto it2 = heap.find(fake_data); + EXPECT_EQ(heap.end(), it2) << + "find by indirection for not included element should fail"; + } + + { + auto it1 = heap.find(Elem(-7)); + EXPECT_NE(heap.end(), it1) << + "find by value for included element should succeed"; + EXPECT_EQ(-7, it1->data) << + "find by value for included element should result in right value"; + + auto it2 = heap.find(Elem(7)); + EXPECT_EQ(heap.end(), it2) << + "find by value for not included element should fail"; + } + + { + auto it1 = heap.rfind(data7); + EXPECT_NE(heap.end(), it1) << + "reverse find by indirecton for included element should succeed"; + EXPECT_EQ(-7, it1->data) << + "reverse find by indirection for included element should result " + "in right value"; + + auto fake_data = std::make_shared<Elem>(-7); + auto it2 = heap.rfind(fake_data); + EXPECT_EQ(heap.end(), it2) << + "reverse find by indirection for not included element should fail"; + } + + { + auto it1 = heap.rfind(Elem(-7)); + EXPECT_NE(heap.end(), it1) << + "reverse find by value for included element should succeed"; + EXPECT_EQ(-7, it1->data) << + "reverse find by value for included element should result " + "in right value"; + + auto it2 = heap.rfind(Elem(7)); + EXPECT_EQ(heap.end(), it2) << + "reverse find by value for not included element should fail"; + } +} + + +TEST_F(HeapFixture1, const_iterator_find_rfind) { + const auto& c_heap = heap; + + { + auto it1 = c_heap.find(data7); + EXPECT_NE(c_heap.cend(), it1) << + "find by indirection for included element should succeed"; + EXPECT_EQ(-7, it1->data) << + "find by indirection for included element should result in right value"; + + auto fake_data = std::make_shared<Elem>(-7); + auto it2 = c_heap.find(fake_data); + EXPECT_EQ(c_heap.cend(), it2) << + "find by indirection for not included element should fail"; + } + + { + auto it1 = c_heap.find(Elem(-7)); + EXPECT_NE(c_heap.cend(), it1) << + "find by value for included element should succeed"; + EXPECT_EQ(-7, it1->data) << + "find by value for included element should result in right value"; + + auto it2 = c_heap.find(Elem(7)); + EXPECT_EQ(c_heap.cend(), it2) << + "find by value for not included element should fail"; + } + + { + auto it1 = c_heap.rfind(data7); + EXPECT_NE(c_heap.cend(), it1) << + "reverse find by indirecton for included element should succeed"; + EXPECT_EQ(-7, it1->data) << + "reverse find by indirection for included element should result " + "in right value"; + + auto fake_data = std::make_shared<Elem>(-7); + auto it2 = c_heap.rfind(fake_data); + EXPECT_EQ(c_heap.cend(), it2) << + "reverse find by indirection for not included element should fail"; + } + + { + auto it1 = c_heap.rfind(Elem(-7)); + EXPECT_NE(c_heap.cend(), it1) << + "reverse find by value for included element should succeed"; + EXPECT_EQ(-7, it1->data) << + "reverse find by value for included element should result " + "in right value"; + + auto it2 = c_heap.rfind(Elem(7)); + EXPECT_EQ(c_heap.cend(), it2) << + "reverse find by value for not included element should fail"; + } +} + + +TEST_F(HeapFixture1, iterator_remove) { + auto it1 = heap.find(data7); + EXPECT_NE(heap.end(), it1) << "find for included element should succeed"; + + heap.remove(it1); + + auto it2 = heap.find(data7); + EXPECT_EQ(heap.end(), it2) << "find for removed element should fail"; + + for (auto it3 = heap.begin(); it3 != heap.end(); ++it3) { + EXPECT_NE(-7, it3->data) << + "iterating through heap should not find removed value"; + } + + // move through heap without -7 + EXPECT_EQ(-12, heap.top().data); + heap.pop(); + EXPECT_EQ(-5, heap.top().data); + heap.pop(); + EXPECT_EQ(1, heap.top().data); + heap.pop(); + EXPECT_EQ(2, heap.top().data); + heap.pop(); + EXPECT_EQ(12, heap.top().data); + heap.pop(); + EXPECT_EQ(99, heap.top().data); + heap.pop(); +} + + +TEST_F(HeapFixture1, four_tops) { + Elem& top1 = heap.top(); + EXPECT_EQ(-12, top1.data); + + const Elem& top2 = heap.top(); + EXPECT_EQ(-12, top2.data); + + std::shared_ptr<Elem> top3 = heap.top_ind(); + EXPECT_EQ(-12, top3->data); + + const std::shared_ptr<Elem> top4 = heap.top_ind(); + EXPECT_EQ(-12, top4->data); + + const auto& c_heap = heap; + + const Elem& top5 = c_heap.top(); + EXPECT_EQ(-12, top5.data); + + const std::shared_ptr<Elem> top6 = c_heap.top_ind(); + EXPECT_EQ(-12, top6->data); +} + + +TEST_F(HeapFixture1, display_sorted) { + std::stringstream ss; + + heap.display_sorted(ss); + + std::string s = ss.str(); + + EXPECT_GT(s.length(), 0u); + + auto negseven = s.find("-7"); + EXPECT_NE(negseven, std::string::npos); + + auto ninetynine = s.find("99"); + EXPECT_NE(ninetynine, std::string::npos); + + // index of -7 should be less than index of 99 + EXPECT_LT(negseven, ninetynine); + +#if 0 + std::cout << s << std::endl; +#endif +} diff --git a/src/dmclock/support/test/test_intrusive_heap.cc b/src/dmclock/support/test/test_intrusive_heap.cc new file mode 100644 index 000000000..18d770f31 --- /dev/null +++ b/src/dmclock/support/test/test_intrusive_heap.cc @@ -0,0 +1,93 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include <string> +#include <iostream> + +#include "intrusive_heap.h" + + +struct TestCompare; +struct TestIntruData; + + +class Test1 { + friend TestCompare; + friend TestIntruData; + + int data; + crimson::IntruHeapData heap_data; + +public: + explicit Test1(int _data) : data(_data) {} + + friend std::ostream& operator<<(std::ostream& out, const Test1& d) { + out << d.data << " (" << d.heap_data << ")"; + return out; + } + + int& the_data() { return data; } +}; + + +struct TestCompare { + bool operator()(const Test1& d1, const Test1& d2) { + return d1.data < d2.data; + } +}; + + +struct TestIntruData { + crimson::IntruHeapData& operator()(Test1& d) { + return d.heap_data; + } +}; + + +int main(int argc, char** argv) { + Test1 d1(2); + Test1 d2(3); + Test1 d3(1); + Test1 d4(-5); + + crimson::IntruHeap<Test1, TestIntruData, TestCompare> my_heap; + + my_heap.push(d1); + my_heap.push(d2); + my_heap.push(d3); + my_heap.push(d4); + my_heap.push(Test1(-9)); + my_heap.push(Test1(99)); + my_heap.push(Test1(0)); + + std::cout << my_heap << std::endl; + + auto& t = my_heap.top(); + t.the_data() = 17; + my_heap.adjust_down(t); + + std::cout << my_heap << std::endl; + + my_heap.display_sorted(std::cout); + + while (!my_heap.empty()) { + auto& top = my_heap.top(); + std::cout << top << std::endl; + my_heap.pop(); + std::cout << my_heap << std::endl; + } + + return 0; +} diff --git a/src/dmclock/test/CMakeLists.txt b/src/dmclock/test/CMakeLists.txt new file mode 100644 index 000000000..52ab6bfea --- /dev/null +++ b/src/dmclock/test/CMakeLists.txt @@ -0,0 +1,34 @@ +include(CheckIncludeFileCXX) +check_include_file_cxx("sys/prctl.h" HAVE_SYS_PRCTL_H) + +set(support_srcs ../sim/src/test_dmclock.cc) +set(test_srcs + test_test_client.cc + test_dmclock_server.cc + test_dmclock_client.cc + ) + +set_source_files_properties(${core_srcs} ${test_srcs} + PROPERTIES + COMPILE_FLAGS "${local_flags}" + ) + +add_executable(dmclock-tests ${test_srcs} ${support_srcs}) +if(HAVE_SYS_PRCTL_H) + target_compile_definitions(dmclock-tests PRIVATE "HAVE_SYS_PRCTL_H") +endif() +target_include_directories(dmclock-tests PRIVATE + ../sim/src ${CMAKE_CURRENT_BINARY_DIR}) +target_include_directories(dmclock-tests PRIVATE SYSTEM + "${GTEST_INCLUDE_DIRS}") + +target_link_libraries(dmclock-tests LINK_PRIVATE + dmclock + Threads::Threads + GTest::GTest + GTest::Main) + +add_test(NAME dmclock-tests + COMMAND $<TARGET_FILE:dmclock-tests>) +add_test(NAME dmclock-data-struct-tests + COMMAND $<TARGET_FILE:dmclock-data-struct-tests>) diff --git a/src/dmclock/test/dmcPrCtl.h b/src/dmclock/test/dmcPrCtl.h new file mode 100644 index 000000000..e0b2216f2 --- /dev/null +++ b/src/dmclock/test/dmcPrCtl.h @@ -0,0 +1,56 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Ceph - scalable distributed file system + * + * Copyright (C) 2017 Red Hat Inc. + * + * This is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License version 2.1, as published by the Free Software + * Foundation. See file COPYING. + * + */ + +// essentially the same as ceph's PrCtl.h, copied into the dmclock library + +#ifdef HAVE_SYS_PRCTL_H +#include <iostream> +#include <sys/prctl.h> +#include <errno.h> + +struct PrCtl { + int saved_state = -1; + int set_dumpable(int new_state) { + int r = prctl(PR_SET_DUMPABLE, new_state); + if (r) { + r = -errno; + std::cerr << "warning: unable to " << (new_state ? "set" : "unset") + << " dumpable flag: " << strerror(r) + << std::endl; + } + return r; + } + PrCtl(int new_state = 0) { + int r = prctl(PR_GET_DUMPABLE); + if (r == -1) { + r = errno; + std::cerr << "warning: unable to get dumpable flag: " << strerror(r) + << std::endl; + } else if (r != new_state) { + if (!set_dumpable(new_state)) { + saved_state = r; + } + } + } + ~PrCtl() { + if (saved_state < 0) { + return; + } + set_dumpable(saved_state); + } +}; +#else +struct PrCtl {}; +#endif diff --git a/src/dmclock/test/test_dmclock_client.cc b/src/dmclock/test/test_dmclock_client.cc new file mode 100644 index 000000000..58840457b --- /dev/null +++ b/src/dmclock/test/test_dmclock_client.cc @@ -0,0 +1,307 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include <chrono> +#include <mutex> +#include <functional> +#include <iostream> + + +#include "dmclock_client.h" +#include "dmclock_util.h" +#include "gtest/gtest.h" + + +namespace dmc = crimson::dmclock; + + +namespace crimson { + namespace dmclock { + + /* + * Allows us to test the code provided with the mutex provided locked. + */ + static void test_locked(std::mutex& mtx, std::function<void()> code) { + std::lock_guard<std::mutex> l(mtx); + code(); + } + + + TEST(dmclock_client, server_erase) { + using ServerId = int; + // using ClientId = int; + + ServerId server = 101; + // ClientId client = 3; + + // dmc::PhaseType resp_params = dmc::PhaseType::reservation; + + dmc::ServiceTracker<ServerId> st(std::chrono::seconds(2), + std::chrono::seconds(3)); + + auto lock_st = [&](std::function<void()> code) { + test_locked(st.data_mtx, code); + }; + + /* The timeline should be as follows: + * + * 0 seconds : request created + * + * 1 seconds : map is size 1 + * + * 2 seconds : clean notes first mark; +2 is base for further calcs + * + * 4 seconds : clean does nothing except makes another mark + * + * 5 seconds : when we're secheduled to erase (+2 + 3) + * + * 5 seconds : since the clean job hasn't run yet, map still size 1 + * + * 6 seconds : clean erases server + * + * 7 seconds : verified server is gone (map size 0) + */ + + lock_st([&] () { + EXPECT_EQ(0u, st.server_map.size()) << + "server map initially has size 0"; + }); + + std::this_thread::sleep_for(std::chrono::seconds(1)); + + // call for side effects + (void) st.get_req_params(server); + + lock_st([&] () { + EXPECT_EQ(1u, st.server_map.size()) << + "server map has size 1 after first request"; + }); + + std::this_thread::sleep_for(std::chrono::seconds(4)); + + lock_st([&] () { + EXPECT_EQ(1u, st.server_map.size()) << + "server map has size 1 just before erase"; + }); + + std::this_thread::sleep_for(std::chrono::seconds(2)); + + lock_st([&] () { + EXPECT_EQ(0u, st.server_map.size()) << + "server map has size 0 just after erase"; + }); + } // TEST + + + TEST(dmclock_client, delta_rho_values_borrowing_tracker) { + using ServerId = int; + // using ClientId = int; + + ServerId server1 = 101; + ServerId server2 = 7; + // ClientId client = 3; + + // RespParams<ServerId> resp_params(server, dmc::PhaseType::reservation); + + dmc::ServiceTracker<ServerId,dmc::BorrowingTracker> st(std::chrono::seconds(2), + std::chrono::seconds(3)); + auto rp1 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp1.delta) << + "delta should be 1 with no intervening responses by" << + "other servers"; + EXPECT_EQ(1u, rp1.rho) << + "rho should be 1 with no intervening reservation responses by" << + "other servers"; + + auto rp2 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp2.delta) << + "delta should be 1 with no intervening responses by" << + "other servers"; + EXPECT_EQ(1u, rp2.rho) << + "rho should be 1 with no intervening reservation responses by" << + "other servers"; + + // RESPONSE + st.track_resp(server1, dmc::PhaseType::priority, 1u); + + auto rp3 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp3.delta) << + "delta should be 1 with no intervening responses by" << + "other servers"; + EXPECT_EQ(1u, rp3.rho) << + "rho should be 1 with no intervening reservation responses by" << + "other servers"; + + // RESPONSE + st.track_resp(server2, dmc::PhaseType::priority, 1u); + + auto rp4 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp4.delta) << + "delta should be 2 with one intervening priority response by " << + "another server"; + EXPECT_EQ(1u, rp4.rho) << + "rho should be 1 with one intervening priority responses by " << + "another server"; + + auto rp5 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp5.delta) << + "delta should be 1 with no intervening responses by" << + "other servers"; + EXPECT_EQ(1u, rp5.rho) << + "rho should be 1 with no intervening reservation responses by" << + "other servers"; + + // RESPONSE + st.track_resp(server2, dmc::PhaseType::reservation, 1u); + + auto rp6 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp6.delta) << + "delta should be 2 with one intervening reservation response by " << + "another server"; + EXPECT_EQ(1u, rp6.rho) << + "rho should be 2 with one intervening reservation responses by " << + "another server"; + + st.track_resp(server2, dmc::PhaseType::reservation, 1u); + st.track_resp(server1, dmc::PhaseType::priority, 1u); + st.track_resp(server2, dmc::PhaseType::priority, 1u); + st.track_resp(server2, dmc::PhaseType::reservation, 1u); + st.track_resp(server1, dmc::PhaseType::reservation, 1u); + st.track_resp(server1, dmc::PhaseType::priority, 1u); + st.track_resp(server2, dmc::PhaseType::priority, 1u); + + auto rp7 = st.get_req_params(server1); + + EXPECT_EQ(5u, rp7.delta) << + "delta should be 5 with four intervening responses by " << + "another server"; + EXPECT_EQ(1u, rp7.rho) << + "rho should be 1 with two intervening reservation responses by " << + "another server"; + + auto rp7b = st.get_req_params(server2); + + EXPECT_EQ(9u, rp7b.delta) << + "delta should be 9 with three intervening responses by " << + "another server"; + EXPECT_EQ(4u, rp7b.rho) << + "rho should be 4 with one intervening reservation responses by " << + "another server"; + + auto rp8 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp8.delta) << + "delta should be 1 with no intervening responses by " << + "another server"; + EXPECT_EQ(1u, rp8.rho) << + "rho should be 1 with no intervening reservation responses by " << + "another server"; + + auto rp8b = st.get_req_params(server2); + EXPECT_EQ(1u, rp8b.delta) << + "delta should be 1 with no intervening responses by " << + "another server"; + EXPECT_EQ(1u, rp8b.rho) << + "rho should be 1 with no intervening reservation responses by " << + "another server"; + } // TEST + + + // NB: the BorrowingTracker has not been fully tested and the + // expected values below have not yet been compared with the + // theoretically correct values. + TEST(dmclock_client, delta_rho_values_orig_tracker) { + using ServerId = int; + + ServerId server1 = 101; + ServerId server2 = 7; + + dmc::ServiceTracker<ServerId,OrigTracker> + st(std::chrono::seconds(2), std::chrono::seconds(3)); + + auto rp1 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp1.delta); + EXPECT_EQ(1u, rp1.rho); + + auto rp2 = st.get_req_params(server1); + + EXPECT_EQ(0u, rp2.delta); + EXPECT_EQ(0u, rp2.rho); + + st.track_resp(server1, dmc::PhaseType::priority, 1u); + + auto rp3 = st.get_req_params(server1); + + EXPECT_EQ(0u, rp3.delta); + EXPECT_EQ(0u, rp3.rho); + + st.track_resp(server2, dmc::PhaseType::priority, 1u); + + auto rp4 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp4.delta); + EXPECT_EQ(0u, rp4.rho); + + auto rp5 = st.get_req_params(server1); + + EXPECT_EQ(0u, rp5.delta); + EXPECT_EQ(0u, rp5.rho); + + st.track_resp(server2, dmc::PhaseType::reservation, 1u); + + auto rp6 = st.get_req_params(server1); + + EXPECT_EQ(1u, rp6.delta); + EXPECT_EQ(1u, rp6.rho); + + // auto rp6_b = st.get_req_params(server2); + + st.track_resp(server2, dmc::PhaseType::reservation, 1u); + st.track_resp(server1, dmc::PhaseType::priority, 1u); + st.track_resp(server2, dmc::PhaseType::priority, 1u); + st.track_resp(server2, dmc::PhaseType::reservation, 1u); + st.track_resp(server1, dmc::PhaseType::reservation, 1u); + st.track_resp(server1, dmc::PhaseType::priority, 1u); + st.track_resp(server2, dmc::PhaseType::priority, 1u); + + auto rp7 = st.get_req_params(server1); + + EXPECT_EQ(4u, rp7.delta); + EXPECT_EQ(2u, rp7.rho); + + auto rp7b = st.get_req_params(server2); + + EXPECT_EQ(3u, rp7b.delta); + EXPECT_EQ(1u, rp7b.rho); + + auto rp8 = st.get_req_params(server1); + + EXPECT_EQ(0u, rp8.delta); + EXPECT_EQ(0u, rp8.rho); + + auto rp8b = st.get_req_params(server2); + EXPECT_EQ(0u, rp8b.delta); + EXPECT_EQ(0u, rp8b.rho); + } // TEST + + } // namespace dmclock +} // namespace crimson diff --git a/src/dmclock/test/test_dmclock_server.cc b/src/dmclock/test/test_dmclock_server.cc new file mode 100644 index 000000000..f591e2e40 --- /dev/null +++ b/src/dmclock/test/test_dmclock_server.cc @@ -0,0 +1,1140 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include <memory> +#include <chrono> +#include <iostream> +#include <list> +#include <vector> + + +#include "dmclock_server.h" +#include "dmclock_util.h" +#include "gtest/gtest.h" + +// process control to prevent core dumps during gtest death tests +#include "dmcPrCtl.h" + + +namespace dmc = crimson::dmclock; + + +// we need a request object; an empty one will do +struct Request { +}; + + +namespace crimson { + namespace dmclock { + + /* + * Allows us to test the code provided with the mutex provided locked. + */ + static void test_locked(std::mutex& mtx, std::function<void()> code) { + std::unique_lock<std::mutex> l(mtx); + code(); + } + + + TEST(dmclock_server, bad_tag_deathtest) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request,true>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 17; + ClientId client2 = 18; + + double reservation = 0.0; + double weight = 0.0; + + dmc::ClientInfo ci1(reservation, weight, 0.0); + dmc::ClientInfo ci2(reservation, weight, 1.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + if (client1 == c) return &ci1; + else if (client2 == c) return &ci2; + else { + ADD_FAILURE() << "got request from neither of two clients"; + return nullptr; + } + }; + + QueueRef pq(new Queue(client_info_f, AtLimit::Wait)); + ReqParams req_params(1,1); + + // Disable coredumps + PrCtl unset_dumpable; + + EXPECT_DEATH_IF_SUPPORTED(pq->add_request(Request{}, client1, req_params), + "Assertion.*reservation.*max_tag.*" + "proportion.*max_tag") << + "we should fail if a client tries to generate a reservation tag " + "where reservation and proportion are both 0"; + + + EXPECT_DEATH_IF_SUPPORTED(pq->add_request(Request{}, client2, req_params), + "Assertion.*reservation.*max_tag.*" + "proportion.*max_tag") << + "we should fail if a client tries to generate a reservation tag " + "where reservation and proportion are both 0"; + + EXPECT_DEATH_IF_SUPPORTED(Queue(client_info_f, AtLimit::Reject), + "Assertion.*Reject.*Delayed") << + "we should fail if a client tries to construct a queue with both " + "DelayedTagCalc and AtLimit::Reject"; + } + + + TEST(dmclock_server, client_idle_erase) { + using ClientId = int; + using Queue = dmc::PushPriorityQueue<ClientId,Request>; + ClientId client = 17; + double reservation = 100.0; + + dmc::ClientInfo ci(reservation, 1.0, 0.0); + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &ci; + }; + auto server_ready_f = [] () -> bool { return true; }; + auto submit_req_f = [] (const ClientId& c, + std::unique_ptr<Request> req, + dmc::PhaseType phase, + uint64_t req_cost) { + // empty; do nothing + }; + + Queue pq(client_info_f, + server_ready_f, + submit_req_f, + std::chrono::seconds(3), + std::chrono::seconds(5), + std::chrono::seconds(2), + AtLimit::Wait); + + auto lock_pq = [&](std::function<void()> code) { + test_locked(pq.data_mtx, code); + }; + + + /* The timeline should be as follows: + * + * 0 seconds : request created + * + * 1 seconds : map is size 1, idle is false + * + * 2 seconds : clean notes first mark; +2 is base for further calcs + * + * 4 seconds : clean does nothing except makes another mark + * + * 5 seconds : when we're secheduled to idle (+2 + 3) + * + * 6 seconds : clean idles client + * + * 7 seconds : when we're secheduled to erase (+2 + 5) + * + * 7 seconds : verified client is idle + * + * 8 seconds : clean erases client info + * + * 9 seconds : verified client is erased + */ + + lock_pq([&] () { + EXPECT_EQ(0u, pq.client_map.size()) << + "client map initially has size 0"; + }); + + Request req; + dmc::ReqParams req_params(1, 1); + EXPECT_EQ(0, pq.add_request_time(req, client, req_params, dmc::get_time())); + + std::this_thread::sleep_for(std::chrono::seconds(1)); + + lock_pq([&] () { + EXPECT_EQ(1u, pq.client_map.size()) << + "client map has 1 after 1 client"; + EXPECT_FALSE(pq.client_map.at(client)->idle) << + "initially client map entry shows not idle."; + }); + + std::this_thread::sleep_for(std::chrono::seconds(6)); + + lock_pq([&] () { + EXPECT_TRUE(pq.client_map.at(client)->idle) << + "after idle age client map entry shows idle."; + }); + + std::this_thread::sleep_for(std::chrono::seconds(2)); + + lock_pq([&] () { + EXPECT_EQ(0u, pq.client_map.size()) << + "client map loses its entry after erase age"; + }); + } // TEST + + + TEST(dmclock_server, delayed_tag_calc) { + using ClientId = int; + constexpr ClientId client1 = 17; + + using DelayedQueue = PullPriorityQueue<ClientId, Request, true>; + using ImmediateQueue = PullPriorityQueue<ClientId, Request, false>; + + ClientInfo info(0.0, 1.0, 1.0); + auto client_info_f = [&] (ClientId c) -> const ClientInfo* { + return &info; + }; + + Time t{1}; + { + DelayedQueue queue(client_info_f); + + queue.add_request_time({}, client1, {0,0}, t); + queue.add_request_time({}, client1, {0,0}, t + 1); + queue.add_request_time({}, client1, {10,10}, t + 2); + + auto pr1 = queue.pull_request(t); + ASSERT_TRUE(pr1.is_retn()); + auto pr2 = queue.pull_request(t + 1); + // ReqParams{10,10} from request #3 pushes request #2 over limit by 10s + ASSERT_TRUE(pr2.is_future()); + EXPECT_DOUBLE_EQ(t + 11, pr2.getTime()); + } + { + ImmediateQueue queue(client_info_f); + + queue.add_request_time({}, client1, {0,0}, t); + queue.add_request_time({}, client1, {0,0}, t + 1); + queue.add_request_time({}, client1, {10,10}, t + 2); + + auto pr1 = queue.pull_request(t); + ASSERT_TRUE(pr1.is_retn()); + auto pr2 = queue.pull_request(t + 1); + // ReqParams{10,10} from request #3 has no effect on request #2 + ASSERT_TRUE(pr2.is_retn()); + auto pr3 = queue.pull_request(t + 2); + ASSERT_TRUE(pr3.is_future()); + EXPECT_DOUBLE_EQ(t + 12, pr3.getTime()); + } + } + +#if 0 + TEST(dmclock_server, reservation_timing) { + using ClientId = int; + // NB? PUSH OR PULL + using Queue = std::unique_ptr<dmc::PriorityQueue<ClientId,Request>>; + using std::chrono::steady_clock; + + int client = 17; + + std::vector<dmc::Time> times; + std::mutex times_mtx; + using Guard = std::lock_guard<decltype(times_mtx)>; + + // reservation every second + dmc::ClientInfo ci(1.0, 0.0, 0.0); + Queue pq; + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &ci; + }; + auto server_ready_f = [] () -> bool { return true; }; + auto submit_req_f = [&] (const ClientId& c, + std::unique_ptr<Request> req, + dmc::PhaseType phase) { + { + Guard g(times_mtx); + times.emplace_back(dmc::get_time()); + } + std::thread complete([&](){ pq->request_completed(); }); + complete.detach(); + }; + + // NB? PUSH OR PULL + pq = Queue(new dmc::PriorityQueue<ClientId,Request>(client_info_f, + server_ready_f, + submit_req_f, + false)); + + Request req; + ReqParams<ClientId> req_params(client, 1, 1); + + for (int i = 0; i < 5; ++i) { + pq->add_request_time(req, req_params, dmc::get_time()); + } + + { + Guard g(times_mtx); + std::this_thread::sleep_for(std::chrono::milliseconds(5500)); + EXPECT_EQ(5, times.size()) << + "after 5.5 seconds, we should have 5 requests times at 1 second apart"; + } + } // TEST +#endif + + + TEST(dmclock_server, remove_by_req_filter) { + struct MyReq { + int id; + + MyReq(int _id) : + id(_id) + { + // empty + } + }; // MyReq + + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,MyReq>; + using MyReqRef = typename Queue::RequestRef; + + ClientId client1 = 17; + ClientId client2 = 98; + + dmc::ClientInfo info1(0.0, 1.0, 0.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info1; + }; + + Queue pq(client_info_f, AtLimit::Allow); + + EXPECT_EQ(0u, pq.client_count()); + EXPECT_EQ(0u, pq.request_count()); + + ReqParams req_params(1,1); + + EXPECT_EQ(0, pq.add_request(MyReq(1), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(11), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(2), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(0), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(13), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(2), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(13), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(98), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(44), client1, req_params)); + + EXPECT_EQ(2u, pq.client_count()); + EXPECT_EQ(9u, pq.request_count()); + + pq.remove_by_req_filter([](MyReqRef&& r) -> bool {return 1 == r->id % 2;}); + + EXPECT_EQ(5u, pq.request_count()); + + std::list<MyReq> capture; + pq.remove_by_req_filter( + [&capture] (MyReqRef&& r) -> bool { + if (0 == r->id % 2) { + capture.push_front(*r); + return true; + } else { + return false; + } + }, + true); + + EXPECT_EQ(0u, pq.request_count()); + EXPECT_EQ(5u, capture.size()); + int total = 0; + for (auto i : capture) { + total += i.id; + } + EXPECT_EQ(146, total) << " sum of captured items should be 146"; + } // TEST + + + TEST(dmclock_server, remove_by_req_filter_ordering_forwards_visit) { + struct MyReq { + int id; + + MyReq(int _id) : + id(_id) + { + // empty + } + }; // MyReq + + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,MyReq>; + using MyReqRef = typename Queue::RequestRef; + + ClientId client1 = 17; + + dmc::ClientInfo info1(0.0, 1.0, 0.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info1; + }; + + Queue pq(client_info_f, AtLimit::Allow); + + EXPECT_EQ(0u, pq.client_count()); + EXPECT_EQ(0u, pq.request_count()); + + ReqParams req_params(1,1); + + EXPECT_EQ(0, pq.add_request(MyReq(1), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(2), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(3), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(4), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(5), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(6), client1, req_params)); + + EXPECT_EQ(1u, pq.client_count()); + EXPECT_EQ(6u, pq.request_count()); + + // remove odd ids in forward order and append to end + + std::vector<MyReq> capture; + pq.remove_by_req_filter( + [&capture] (MyReqRef&& r) -> bool { + if (1 == r->id % 2) { + capture.push_back(*r); + return true; + } else { + return false; + } + }, + false); + + EXPECT_EQ(3u, pq.request_count()); + EXPECT_EQ(3u, capture.size()); + EXPECT_EQ(1, capture[0].id) << "items should come out in forward order"; + EXPECT_EQ(3, capture[1].id) << "items should come out in forward order"; + EXPECT_EQ(5, capture[2].id) << "items should come out in forward order"; + + // remove even ids in reverse order but insert at front so comes + // out forwards + + std::vector<MyReq> capture2; + pq.remove_by_req_filter( + [&capture2] (MyReqRef&& r) -> bool { + if (0 == r->id % 2) { + capture2.insert(capture2.begin(), *r); + return true; + } else { + return false; + } + }, + false); + + EXPECT_EQ(0u, pq.request_count()); + EXPECT_EQ(3u, capture2.size()); + EXPECT_EQ(6, capture2[0].id) << "items should come out in reverse order"; + EXPECT_EQ(4, capture2[1].id) << "items should come out in reverse order"; + EXPECT_EQ(2, capture2[2].id) << "items should come out in reverse order"; + } // TEST + + + TEST(dmclock_server, remove_by_req_filter_ordering_backwards_visit) { + struct MyReq { + int id; + + MyReq(int _id) : + id(_id) + { + // empty + } + }; // MyReq + + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,MyReq>; + using MyReqRef = typename Queue::RequestRef; + + ClientId client1 = 17; + + dmc::ClientInfo info1(0.0, 1.0, 0.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info1; + }; + + Queue pq(client_info_f, AtLimit::Allow); + + EXPECT_EQ(0u, pq.client_count()); + EXPECT_EQ(0u, pq.request_count()); + + ReqParams req_params(1,1); + + EXPECT_EQ(0, pq.add_request(MyReq(1), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(2), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(3), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(4), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(5), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(6), client1, req_params)); + + EXPECT_EQ(1u, pq.client_count()); + EXPECT_EQ(6u, pq.request_count()); + + // now remove odd ids in forward order + + std::vector<MyReq> capture; + pq.remove_by_req_filter( + [&capture] (MyReqRef&& r) -> bool { + if (1 == r->id % 2) { + capture.insert(capture.begin(), *r); + return true; + } else { + return false; + } + }, + true); + + EXPECT_EQ(3u, pq.request_count()); + EXPECT_EQ(3u, capture.size()); + EXPECT_EQ(1, capture[0].id) << "items should come out in forward order"; + EXPECT_EQ(3, capture[1].id) << "items should come out in forward order"; + EXPECT_EQ(5, capture[2].id) << "items should come out in forward order"; + + // now remove even ids in reverse order + + std::vector<MyReq> capture2; + pq.remove_by_req_filter( + [&capture2] (MyReqRef&& r) -> bool { + if (0 == r->id % 2) { + capture2.push_back(*r); + return true; + } else { + return false; + } + }, + true); + + EXPECT_EQ(0u, pq.request_count()); + EXPECT_EQ(3u, capture2.size()); + EXPECT_EQ(6, capture2[0].id) << "items should come out in reverse order"; + EXPECT_EQ(4, capture2[1].id) << "items should come out in reverse order"; + EXPECT_EQ(2, capture2[2].id) << "items should come out in reverse order"; + } // TEST + + + TEST(dmclock_server, remove_by_client) { + struct MyReq { + int id; + + MyReq(int _id) : + id(_id) + { + // empty + } + }; // MyReq + + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,MyReq>; + using MyReqRef = typename Queue::RequestRef; + + ClientId client1 = 17; + ClientId client2 = 98; + + dmc::ClientInfo info1(0.0, 1.0, 0.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info1; + }; + + Queue pq(client_info_f, AtLimit::Allow); + + EXPECT_EQ(0u, pq.client_count()); + EXPECT_EQ(0u, pq.request_count()); + + ReqParams req_params(1,1); + + EXPECT_EQ(0, pq.add_request(MyReq(1), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(11), client1, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(2), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(0), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(13), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(2), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(13), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(98), client2, req_params)); + EXPECT_EQ(0, pq.add_request(MyReq(44), client1, req_params)); + + EXPECT_EQ(2u, pq.client_count()); + EXPECT_EQ(9u, pq.request_count()); + + std::list<MyReq> removed; + + pq.remove_by_client(client1, + true, + [&removed] (MyReqRef&& r) { + removed.push_front(*r); + }); + + EXPECT_EQ(3u, removed.size()); + EXPECT_EQ(1, removed.front().id); + removed.pop_front(); + EXPECT_EQ(11, removed.front().id); + removed.pop_front(); + EXPECT_EQ(44, removed.front().id); + removed.pop_front(); + + EXPECT_EQ(6u, pq.request_count()); + + Queue::PullReq pr = pq.pull_request(); + EXPECT_TRUE(pr.is_retn()); + EXPECT_EQ(2, pr.get_retn().request->id); + + pr = pq.pull_request(); + EXPECT_TRUE(pr.is_retn()); + EXPECT_EQ(0, pr.get_retn().request->id); + + pq.remove_by_client(client2); + EXPECT_EQ(0u, pq.request_count()) << + "after second client removed, none left"; + } // TEST + + + TEST(dmclock_server_pull, pull_weight) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 17; + ClientId client2 = 98; + + dmc::ClientInfo info1(0.0, 1.0, 0.0); + dmc::ClientInfo info2(0.0, 2.0, 0.0); + + QueueRef pq; + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + if (client1 == c) return &info1; + else if (client2 == c) return &info2; + else { + ADD_FAILURE() << "client info looked up for non-existent client"; + return nullptr; + } + }; + + pq = QueueRef(new Queue(client_info_f, AtLimit::Wait)); + + ReqParams req_params(1,1); + + auto now = dmc::get_time(); + + for (int i = 0; i < 5; ++i) { + EXPECT_EQ(0, pq->add_request(Request{}, client1, req_params)); + EXPECT_EQ(0, pq->add_request(Request{}, client2, req_params)); + now += 0.0001; + } + + int c1_count = 0; + int c2_count = 0; + for (int i = 0; i < 6; ++i) { + Queue::PullReq pr = pq->pull_request(); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + auto& retn = boost::get<Queue::PullReq::Retn>(pr.data); + + if (client1 == retn.client) ++c1_count; + else if (client2 == retn.client) ++c2_count; + else ADD_FAILURE() << "got request from neither of two clients"; + + EXPECT_EQ(PhaseType::priority, retn.phase); + } + + EXPECT_EQ(2, c1_count) << + "one-third of request should have come from first client"; + EXPECT_EQ(4, c2_count) << + "two-thirds of request should have come from second client"; + } + + + TEST(dmclock_server_pull, pull_reservation) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 52; + ClientId client2 = 8; + + dmc::ClientInfo info1(2.0, 0.0, 0.0); + dmc::ClientInfo info2(1.0, 0.0, 0.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + if (client1 == c) return &info1; + else if (client2 == c) return &info2; + else { + ADD_FAILURE() << "client info looked up for non-existent client"; + return nullptr; + } + }; + + QueueRef pq(new Queue(client_info_f, AtLimit::Wait)); + + ReqParams req_params(1,1); + + // make sure all times are well before now + auto old_time = dmc::get_time() - 100.0; + + for (int i = 0; i < 5; ++i) { + EXPECT_EQ(0, pq->add_request_time(Request{}, client1, req_params, old_time)); + EXPECT_EQ(0, pq->add_request_time(Request{}, client2, req_params, old_time)); + old_time += 0.001; + } + + int c1_count = 0; + int c2_count = 0; + + for (int i = 0; i < 6; ++i) { + Queue::PullReq pr = pq->pull_request(); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + auto& retn = boost::get<Queue::PullReq::Retn>(pr.data); + + if (client1 == retn.client) ++c1_count; + else if (client2 == retn.client) ++c2_count; + else ADD_FAILURE() << "got request from neither of two clients"; + + EXPECT_EQ(PhaseType::reservation, retn.phase); + } + + EXPECT_EQ(4, c1_count) << + "two-thirds of request should have come from first client"; + EXPECT_EQ(2, c2_count) << + "one-third of request should have come from second client"; + } // dmclock_server_pull.pull_reservation + + + TEST(dmclock_server_pull, update_client_info) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request,false>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 17; + ClientId client2 = 98; + + dmc::ClientInfo info1(0.0, 100.0, 0.0); + dmc::ClientInfo info2(0.0, 200.0, 0.0); + + QueueRef pq; + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + if (client1 == c) return &info1; + else if (client2 == c) return &info2; + else { + ADD_FAILURE() << "client info looked up for non-existent client"; + return nullptr; + } + }; + + pq = QueueRef(new Queue(client_info_f, AtLimit::Wait)); + + ReqParams req_params(1,1); + + auto now = dmc::get_time(); + + for (int i = 0; i < 5; ++i) { + EXPECT_EQ(0, pq->add_request(Request{}, client1, req_params)); + EXPECT_EQ(0, pq->add_request(Request{}, client2, req_params)); + now += 0.0001; + } + + int c1_count = 0; + int c2_count = 0; + for (int i = 0; i < 10; ++i) { + Queue::PullReq pr = pq->pull_request(); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + auto& retn = boost::get<Queue::PullReq::Retn>(pr.data); + + if (i > 5) continue; + if (client1 == retn.client) ++c1_count; + else if (client2 == retn.client) ++c2_count; + else ADD_FAILURE() << "got request from neither of two clients"; + + EXPECT_EQ(PhaseType::priority, retn.phase); + } + + EXPECT_EQ(2, c1_count) << + "before: one-third of request should have come from first client"; + EXPECT_EQ(4, c2_count) << + "before: two-thirds of request should have come from second client"; + + std::chrono::seconds dura(1); + std::this_thread::sleep_for(dura); + + info1 = dmc::ClientInfo(0.0, 200.0, 0.0); + pq->update_client_info(17); + + now = dmc::get_time(); + + for (int i = 0; i < 5; ++i) { + EXPECT_EQ(0, pq->add_request(Request{}, client1, req_params)); + EXPECT_EQ(0, pq->add_request(Request{}, client2, req_params)); + now += 0.0001; + } + + c1_count = 0; + c2_count = 0; + for (int i = 0; i < 6; ++i) { + Queue::PullReq pr = pq->pull_request(); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + auto& retn = boost::get<Queue::PullReq::Retn>(pr.data); + + if (client1 == retn.client) ++c1_count; + else if (client2 == retn.client) ++c2_count; + else ADD_FAILURE() << "got request from neither of two clients"; + + EXPECT_EQ(PhaseType::priority, retn.phase); + } + + EXPECT_EQ(3, c1_count) << + "after: one-third of request should have come from first client"; + EXPECT_EQ(3, c2_count) << + "after: two-thirds of request should have come from second client"; + } + + + TEST(dmclock_server_pull, dynamic_cli_info_f) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request,true,true>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 17; + ClientId client2 = 98; + + std::vector<dmc::ClientInfo> info1; + std::vector<dmc::ClientInfo> info2; + + info1.push_back(dmc::ClientInfo(0.0, 100.0, 0.0)); + info1.push_back(dmc::ClientInfo(0.0, 150.0, 0.0)); + + info2.push_back(dmc::ClientInfo(0.0, 200.0, 0.0)); + info2.push_back(dmc::ClientInfo(0.0, 50.0, 0.0)); + + size_t cli_info_group = 0; + + QueueRef pq; + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + if (client1 == c) return &info1[cli_info_group]; + else if (client2 == c) return &info2[cli_info_group]; + else { + ADD_FAILURE() << "client info looked up for non-existent client"; + return nullptr; + } + }; + + pq = QueueRef(new Queue(client_info_f, AtLimit::Wait)); + + ReqParams req_params(1,1); + + auto now = dmc::get_time(); + + for (int i = 0; i < 5; ++i) { + EXPECT_EQ(0, pq->add_request(Request{}, client1, req_params)); + EXPECT_EQ(0, pq->add_request(Request{}, client2, req_params)); + now += 0.0001; + } + + int c1_count = 0; + int c2_count = 0; + for (int i = 0; i < 10; ++i) { + Queue::PullReq pr = pq->pull_request(); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + auto& retn = boost::get<Queue::PullReq::Retn>(pr.data); + + if (i > 5) continue; + if (client1 == retn.client) ++c1_count; + else if (client2 == retn.client) ++c2_count; + else ADD_FAILURE() << "got request from neither of two clients"; + + EXPECT_EQ(PhaseType::priority, retn.phase); + } + + EXPECT_EQ(2, c1_count) << + "before: one-third of request should have come from first client"; + EXPECT_EQ(4, c2_count) << + "before: two-thirds of request should have come from second client"; + + std::chrono::seconds dura(1); + std::this_thread::sleep_for(dura); + + cli_info_group = 1; + + now = dmc::get_time(); + + for (int i = 0; i < 6; ++i) { + EXPECT_EQ(0, pq->add_request(Request{}, client1, req_params)); + EXPECT_EQ(0, pq->add_request(Request{}, client2, req_params)); + now += 0.0001; + } + + c1_count = 0; + c2_count = 0; + for (int i = 0; i < 8; ++i) { + Queue::PullReq pr = pq->pull_request(); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + auto& retn = boost::get<Queue::PullReq::Retn>(pr.data); + + if (client1 == retn.client) ++c1_count; + else if (client2 == retn.client) ++c2_count; + else ADD_FAILURE() << "got request from neither of two clients"; + + EXPECT_EQ(PhaseType::priority, retn.phase); + } + + EXPECT_EQ(6, c1_count) << + "after: one-third of request should have come from first client"; + EXPECT_EQ(2, c2_count) << + "after: two-thirds of request should have come from second client"; + } + + + // This test shows what happens when a request can be ready (under + // limit) but not schedulable since proportion tag is 0. We expect + // to get some future and none responses. + TEST(dmclock_server_pull, ready_and_under_limit) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 52; + ClientId client2 = 8; + + dmc::ClientInfo info1(1.0, 0.0, 0.0); + dmc::ClientInfo info2(1.0, 0.0, 0.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + if (client1 == c) return &info1; + else if (client2 == c) return &info2; + else { + ADD_FAILURE() << "client info looked up for non-existent client"; + return nullptr; + } + }; + + QueueRef pq(new Queue(client_info_f, AtLimit::Wait)); + + ReqParams req_params(0, 0); + + // make sure all times are well before now + auto start_time = dmc::get_time() - 100.0; + + // add six requests; for same client reservations spaced one apart + for (int i = 0; i < 3; ++i) { + EXPECT_EQ(0, pq->add_request_time(Request{}, client1, req_params, start_time)); + EXPECT_EQ(0, pq->add_request_time(Request{}, client2, req_params, start_time)); + } + + Queue::PullReq pr = pq->pull_request(start_time + 0.5); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + + pr = pq->pull_request(start_time + 0.5); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + + pr = pq->pull_request(start_time + 0.5); + EXPECT_EQ(Queue::NextReqType::future, pr.type) << + "too soon for next reservation"; + + pr = pq->pull_request(start_time + 1.5); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + + pr = pq->pull_request(start_time + 1.5); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + + pr = pq->pull_request(start_time + 1.5); + EXPECT_EQ(Queue::NextReqType::future, pr.type) << + "too soon for next reservation"; + + pr = pq->pull_request(start_time + 2.5); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + + pr = pq->pull_request(start_time + 2.5); + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + + pr = pq->pull_request(start_time + 2.5); + EXPECT_EQ(Queue::NextReqType::none, pr.type) << "no more requests left"; + } + + + TEST(dmclock_server_pull, pull_none) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request>; + using QueueRef = std::unique_ptr<Queue>; + + dmc::ClientInfo info(1.0, 1.0, 1.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info; + }; + + QueueRef pq(new Queue(client_info_f, AtLimit::Wait)); + + // Request req; + ReqParams req_params(1,1); + + auto now = dmc::get_time(); + + Queue::PullReq pr = pq->pull_request(now + 100); + + EXPECT_EQ(Queue::NextReqType::none, pr.type); + } + + + TEST(dmclock_server_pull, pull_future) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 52; + // ClientId client2 = 8; + + dmc::ClientInfo info(1.0, 0.0, 1.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info; + }; + + QueueRef pq(new Queue(client_info_f, AtLimit::Wait)); + + ReqParams req_params(1,1); + + // make sure all times are well before now + auto now = dmc::get_time(); + + EXPECT_EQ(0, pq->add_request_time(Request{}, client1, req_params, now + 100)); + Queue::PullReq pr = pq->pull_request(now); + + EXPECT_EQ(Queue::NextReqType::future, pr.type); + + Time when = boost::get<Time>(pr.data); + EXPECT_EQ(now + 100, when); + } + + + TEST(dmclock_server_pull, pull_future_limit_break_weight) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 52; + // ClientId client2 = 8; + + dmc::ClientInfo info(0.0, 1.0, 1.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info; + }; + + QueueRef pq(new Queue(client_info_f, AtLimit::Allow)); + + ReqParams req_params(1,1); + + // make sure all times are well before now + auto now = dmc::get_time(); + + EXPECT_EQ(0, pq->add_request_time(Request{}, client1, req_params, now + 100)); + Queue::PullReq pr = pq->pull_request(now); + + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + + auto& retn = boost::get<Queue::PullReq::Retn>(pr.data); + EXPECT_EQ(client1, retn.client); + } + + + TEST(dmclock_server_pull, pull_future_limit_break_reservation) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId,Request>; + using QueueRef = std::unique_ptr<Queue>; + + ClientId client1 = 52; + // ClientId client2 = 8; + + dmc::ClientInfo info(1.0, 0.0, 1.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info; + }; + + QueueRef pq(new Queue(client_info_f, AtLimit::Allow)); + + ReqParams req_params(1,1); + + // make sure all times are well before now + auto now = dmc::get_time(); + + EXPECT_EQ(0, pq->add_request_time(Request{}, client1, req_params, now + 100)); + Queue::PullReq pr = pq->pull_request(now); + + EXPECT_EQ(Queue::NextReqType::returning, pr.type); + + auto& retn = boost::get<Queue::PullReq::Retn>(pr.data); + EXPECT_EQ(client1, retn.client); + } + + + TEST(dmclock_server_pull, pull_reject_at_limit) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId, Request, false>; + using MyReqRef = typename Queue::RequestRef; + + ClientId client1 = 52; + ClientId client2 = 53; + + dmc::ClientInfo info(0.0, 1.0, 1.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info; + }; + + Queue pq(client_info_f, AtLimit::Reject); + + { + // success at 1 request per second + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{1})); + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{2})); + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{3})); + // request too soon + EXPECT_EQ(EAGAIN, pq.add_request_time({}, client1, {}, Time{3.9})); + // previous rejected request counts against limit + EXPECT_EQ(EAGAIN, pq.add_request_time({}, client1, {}, Time{4})); + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{6})); + } + { + auto r1 = MyReqRef{new Request}; + ASSERT_EQ(0, pq.add_request(std::move(r1), client2, {}, Time{1})); + EXPECT_EQ(nullptr, r1); // add_request takes r1 on success + auto r2 = MyReqRef{new Request}; + ASSERT_EQ(EAGAIN, pq.add_request(std::move(r2), client2, {}, Time{1})); + EXPECT_NE(nullptr, r2); // add_request does not take r2 on failure + } + } + + + TEST(dmclock_server_pull, pull_reject_threshold) { + using ClientId = int; + using Queue = dmc::PullPriorityQueue<ClientId, Request, false>; + + ClientId client1 = 52; + + dmc::ClientInfo info(0.0, 1.0, 1.0); + + auto client_info_f = [&] (ClientId c) -> const dmc::ClientInfo* { + return &info; + }; + + // allow up to 3 seconds worth of limit before rejecting + Queue pq(client_info_f, RejectThreshold{3.0}); + + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{1})); // at limit=1 + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{1})); // 1 over + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{1})); // 2 over + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{1})); // 3 over + EXPECT_EQ(EAGAIN, pq.add_request_time({}, client1, {}, Time{1})); // reject + EXPECT_EQ(0, pq.add_request_time({}, client1, {}, Time{3})); // 3 over + } + + } // namespace dmclock +} // namespace crimson diff --git a/src/dmclock/test/test_test_client.cc b/src/dmclock/test/test_test_client.cc new file mode 100644 index 000000000..11cbd74b1 --- /dev/null +++ b/src/dmclock/test/test_test_client.cc @@ -0,0 +1,138 @@ +// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*- +// vim: ts=8 sw=2 smarttab + +/* + * Copyright (C) 2016 Red Hat Inc. + * + * Author: J. Eric Ivancich <ivancich@redhat.com> + * + * This is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License version + * 2.1, as published by the Free Software Foundation. See file + * COPYING. + */ + + +#include <atomic> +#include <thread> +#include <chrono> +#include <iostream> + +#include "gtest/gtest.h" + +#include "sim_recs.h" +#include "sim_client.h" + +#include "test_dmclock.h" + + +using namespace std::placeholders; + +namespace dmc = crimson::dmclock; +namespace test = crimson::test_dmc; +namespace sim = crimson::qos_simulation; + +using TimePoint = std::chrono::time_point<std::chrono::system_clock>; + +static TimePoint now() { return std::chrono::system_clock::now(); } + + +TEST(test_client, full_bore_timing) { + std::atomic_ulong count(0); + + ServerId server_id = 3; + + sim::TestResponse resp(0); + dmc::PhaseType resp_params = dmc::PhaseType::priority; + test::DmcClient* client; + const sim::Cost request_cost = 1u; + + auto start = now(); + client = + new test::DmcClient(ClientId(0), + [&] (const ServerId& server, + const sim::TestRequest& req, + const ClientId& client_id, + const dmc::ReqParams& req_params) { + ++count; + client->receive_response(resp, client_id, resp_params, request_cost); + }, + [&] (const uint64_t seed) -> ServerId& { + return server_id; + }, + test::dmc_client_accumulate_f, + 1000, // ops to run + 100, // iops goal + 5); // outstanding ops allowed + client->wait_until_done(); + auto end = now(); + EXPECT_EQ(1000u, count) << "didn't get right number of ops"; + + int milliseconds = (end - start) / std::chrono::milliseconds(1); + EXPECT_LT(10000, milliseconds) << "timing too fast to be correct"; + EXPECT_GT(12000, milliseconds) << "timing suspiciously slow"; + + delete client; +} + + +TEST(test_client, paused_timing) { + std::atomic_ulong count(0); + std::atomic_ulong unresponded_count(0); + std::atomic_bool auto_respond(false); + + ClientId my_client_id = 0; + ServerId server_id = 3; + + sim::TestResponse resp(0); + dmc::PhaseType resp_params = dmc::PhaseType::priority; + const uint64_t request_cost = 1u; + test::DmcClient* client; + + auto start = now(); + client = + new test::DmcClient(my_client_id, + [&] (const ServerId& server, + const sim::TestRequest& req, + const ClientId& client_id, + const dmc::ReqParams& req_params) { + ++count; + if (auto_respond.load()) { + client->receive_response(resp, client_id, resp_params, request_cost); + } else { + ++unresponded_count; + } + }, + [&] (const uint64_t seed) -> ServerId& { + return server_id; + }, + test::dmc_client_accumulate_f, + + 1000, // ops to run + 100, // iops goal + 50); // outstanding ops allowed + std::thread t([&]() { + std::this_thread::sleep_for(std::chrono::seconds(5)); + EXPECT_EQ(50u, unresponded_count.load()) << + "should have 50 unresponded calls"; + auto_respond = true; + // respond to those 50 calls + for(int i = 0; i < 50; ++i) { + client->receive_response(resp, my_client_id, resp_params, 1); + --unresponded_count; + } + }); + + client->wait_until_done(); + auto end = now(); + int milliseconds = (end - start) / std::chrono::milliseconds(1); + + // the 50 outstanding ops allowed means the first half-second of + // requests get responded to during the 5 second pause. So we have + // to adjust our expectations by a half-second. + EXPECT_LT(15000 - 500, milliseconds) << "timing too fast to be correct"; + EXPECT_GT(17000 - 500, milliseconds) << "timing suspiciously slow"; + t.join(); + + delete client; +} |