Spec URL: https://music.fedorapeople.org/atomic-queue.spec SRPM URL: https://music.fedorapeople.org/atomic-queue-0-0.1.20210729git8fec762.fc34.src.rpm Description: C++14 multiple-producer-multiple-consumer lockless queues based on circular buffer with std::atomic. The main design principle these queues follow is minimalism: the bare minimum of atomic operations, fixed size buffer, value semantics. These qualities are also limitations: • The maximum queue size must be set at compile time or construction time. The circular buffer side-steps the memory reclamation problem inherent in linked-list based queues for the price of fixed buffer size. See Effective memory reclamation for lock-free data structures in C++ for more details. Fixed buffer size may not be that much of a limitation, since once the queue gets larger than the maximum expected size that indicates a problem that elements aren’t processed fast enough, and if the queue keeps growing it may eventually consume all available memory which may affect the entire system, rather than the problematic process only. The only apparent inconvenience is that one has to do an upfront back-of-the-envelope calculation on what would be the largest expected/acceptable queue size. • There are no OS-blocking push/pop functions. This queue is designed for ultra-low-latency scenarios and using an OS blocking primitive would be sacrificing push-to-pop latency. For lowest possible latency one cannot afford blocking in the OS kernel because the wake-up latency of a blocked thread is about 1-3 microseconds, whereas this queue’s round-trip time can be as low as 150 nanoseconds. Ultra-low-latency applications need just that and nothing more. The minimalism pays off, see the throughput and latency benchmarks. Available containers are: • AtomicQueue - a fixed size ring-buffer for atomic elements. • OptimistAtomicQueue - a faster fixed size ring-buffer for atomic elements which busy-waits when empty or full. • AtomicQueue2 - a fixed size ring-buffer for non-atomic elements. • OptimistAtomicQueue2 - a faster fixed size ring-buffer for non-atomic elements which busy-waits when empty or full. These containers have corresponding AtomicQueueB, OptimistAtomicQueueB, AtomicQueueB2, OptimistAtomicQueueB2 versions where the buffer size is specified as an argument to the constructor. Totally ordered mode is supported. In this mode consumers receive messages in the same FIFO order the messages were posted. This mode is supported for push and pop functions, but for not the try_ versions. On Intel x86 the totally ordered mode has 0 cost, as of 2019. Single-producer-single-consumer mode is supported. In this mode, no read-modify-write instructions are necessary, only the atomic loads and stores. That improves queue throughput significantly. Fedora Account System Username: music Koji builds: F35: https://koji.fedoraproject.org/koji/taskinfo?taskID=72932278 F34: https://koji.fedoraproject.org/koji/taskinfo?taskID=72932294 F33: https://koji.fedoraproject.org/koji/taskinfo?taskID=72932296 I am aware of https://docs.fedoraproject.org/en-US/packaging-guidelines/#_architecture_build_failures and will file bugs blocking PPCTracker and F-ExcludeArch-s390x after import.
My PR to make building the benchmarks optional was merged. Updated submission: Spec URL: https://music.fedorapeople.org/20210731/atomic-queue.spec SRPM URL: https://music.fedorapeople.org/20210731/atomic-queue-0-0.1.20210731gitaa08199.fc34.src.rpm
I will take this review.
Neither issue below warrants blocking the package. This package is APPROVED. Package Review ============== Legend: [x] = Pass, [!] = Fail, [-] = Not applicable, [?] = Not evaluated Issues: ======= - Note the description-line-too-long warning from rpmlint. There is a macro on the indicated line that expands to a fairly long string. - Regarding support for ppc64le and s390x, it looks like the only assembly is in defs.h, used to define the spin_loop_pause() function, right? If that is the case, then it would be correct (although unfriendly to the CPU) to define an empty spin_loop_pause() function, or to expand to whatever the no-op instruction is for each platform. For PowerPC, something like this (largely stolen from the Linux kernel's arch/powerpc/include/asm/vdso/processor.h): namespace atomic_queue { constexpr int CACHE_LINE_SIZE = 128; static inline void spin_loop_pause() noexcept { asm volatile("or 1, 1, 1" ::: "memory"); } } For s390x, something like this (again stolen from the Linux kernel, but in arch/s390/include/asm/processor.h): namespace atomic_queue { constexpr int CACHE_LINE_SIZE = 256; static inline void spin_loop_pause() noexcept { asm volatile("bcr 14,0" ::: "memory"); } } I know you aren't upstream, just saying that if you want to see the ExclusiveArch go away, it may not be too difficult to make that happen. ===== MUST items ===== C/C++: [x]: Package does not contain kernel modules. [x]: Package contains no static executables. [x]: If your application is a C or C++ application you must list a BuildRequires against gcc, gcc-c++ or clang. [x]: Header files in -devel subpackage, if present. [x]: Package does not contain any libtool archives (.la) [x]: Rpath absent or only used for internal libs. Generic: [x]: Package is licensed with an open-source compatible license and meets other legal requirements as defined in the legal section of Packaging Guidelines. [x]: License field in the package spec file matches the actual license. [x]: %build honors applicable compiler flags or justifies otherwise. [x]: Package contains no bundled libraries without FPC exception. [x]: Changelog in prescribed format. [x]: Sources contain only permissible code or content. [-]: Package contains desktop file if it is a GUI application. [x]: Development files must be in a -devel package [x]: Package uses nothing in %doc for runtime. [x]: Package consistently uses macros (instead of hard-coded directory names). [x]: Package is named according to the Package Naming Guidelines. [x]: Package does not generate any conflict. [x]: Package obeys FHS, except libexecdir and /usr/target. [-]: If the package is a rename of another package, proper Obsoletes and Provides are present. [x]: Requires correct, justified where necessary. [x]: Spec file is legible and written in American English. [-]: Package contains systemd file(s) if in need. [-]: Useful -debuginfo package or justification otherwise. [!]: Package is not known to require an ExcludeArch tag. It needs an ExclusiveArch tag. See comment above about that. [-]: Large documentation must go in a -doc subpackage. Large could be size (~1MB) or number of files. Note: Documentation size is 20480 bytes in 1 files. [x]: Package complies to the Packaging Guidelines [x]: Package successfully compiles and builds into binary rpms on at least one supported primary architecture. [x]: Package installs properly. [x]: Rpmlint is run on all rpms the build produces. Note: There are rpmlint messages (see attachment). [x]: If (and only if) the source package includes the text of the license(s) in its own file, then that file, containing the text of the license(s) for the package is included in %license. [x]: Package requires other packages for directories it uses. [x]: Package must own all directories that it creates. [x]: Package does not own files or directories owned by other packages. [x]: Package uses either %{buildroot} or $RPM_BUILD_ROOT [x]: Package does not run rm -rf %{buildroot} (or $RPM_BUILD_ROOT) at the beginning of %install. [x]: Macros in Summary, %description expandable at SRPM build time. [x]: Package does not contain duplicates in %files. [x]: Permissions on files are set properly. [x]: Package must not depend on deprecated() packages. [x]: Package use %makeinstall only when make install DESTDIR=... doesn't work. [x]: Package is named using only allowed ASCII characters. [x]: Package does not use a name that already exists. [x]: Package is not relocatable. [x]: Sources used to build the package match the upstream source, as provided in the spec URL. [x]: Spec file name must match the spec package %{name}, in the format %{name}.spec. [x]: File names are valid UTF-8. [x]: Packages must not store files under /srv, /opt or /usr/local ===== SHOULD items ===== Generic: [-]: If the source package does not include license text(s) as a separate file from upstream, the packager SHOULD query upstream to include it. [x]: Final provides and requires are sane (see attachments). [?]: Package functions as described. [x]: Latest version is packaged. [x]: Package does not include license text files separate from upstream. [-]: Sources are verified with gpgverify first in %prep if upstream publishes signatures. Note: gpgverify is not used. [-]: Description and summary sections in the package spec file contains translations for supported Non-English languages, if available. [x]: %check is present and all tests pass. [x]: Packages should try to preserve timestamps of original installed files. [x]: Reviewer should test that the package builds in mock. [x]: Buildroot is not present [x]: Package has no %clean section with rm -rf %{buildroot} (or $RPM_BUILD_ROOT) [x]: No file requires outside of /etc, /bin, /sbin, /usr/bin, /usr/sbin. [x]: Packager, Vendor, PreReq, Copyright tags should not be in spec file [x]: Sources can be downloaded from URI in Source: tag [x]: SourceX is a working URL. [x]: Package should compile and build into binary rpms on all supported architectures. [x]: Spec use %global instead of %define unless justified. ===== EXTRA items ===== Generic: [x]: Rpmlint is run on all installed packages. Note: There are rpmlint messages (see attachment). [x]: Large data in /usr/share should live in a noarch subpackage if package is arched. [x]: Spec file according to URL is the same as in SRPM. Rpmlint ------- Checking: atomic-queue-devel-0-0.1.20210731gitaa08199.fc36.x86_64.rpm atomic-queue-0-0.1.20210731gitaa08199.fc36.src.rpm atomic-queue-devel.x86_64: W: spelling-error %description -l en_US lockless -> luckless, lock less, lock-less atomic-queue-devel.x86_64: W: spelling-error %description -l en_US aren -> earn, are, arena atomic-queue-devel.x86_64: E: description-line-too-long C The atomic-queue-devel package contains libraries and header files for developing atomic-queue.src: W: spelling-error Summary(en_US) lockless -> luckless, lock less, lock-less atomic-queue.src: W: spelling-error %description -l en_US lockless -> luckless, lock less, lock-less atomic-queue.src: W: spelling-error %description -l en_US aren -> earn, are, arena atomic-queue.src:124: W: macro-in-%changelog %autochangelog 2 packages and 0 specfiles checked; 1 errors, 6 warnings. Rpmlint (installed packages) ---------------------------- rpmlint: 2.0.0 configuration: /usr/lib/python3.10/site-packages/rpmlint/configdefaults.toml /etc/xdg/rpmlint/fedora.toml /etc/xdg/rpmlint/licenses.toml /etc/xdg/rpmlint/scoring.toml /etc/xdg/rpmlint/users-groups.toml /etc/xdg/rpmlint/warn-on-functions.toml checks: 31, packages: 1 atomic-queue-devel.x86_64: E: description-line-too-long The atomic-queue-devel package contains libraries and header files for developing ================= 1 packages and 0 specfiles checked; 1 errors, 0 warnings, 1 badness; has taken 0.0 s ================= Source checksums ---------------- https://github.com/max0x7ba/atomic_queue/archive/aa08199a7a516a561be1685afb644cf99e5b98e9/atomic_queue-aa08199a7a516a561be1685afb644cf99e5b98e9.tar.gz : CHECKSUM(SHA256) this package : a5bab145a2185993d6c81437f263d7bf557bfd8f4d31d4ad3c720136b5402f63 CHECKSUM(SHA256) upstream package : a5bab145a2185993d6c81437f263d7bf557bfd8f4d31d4ad3c720136b5402f63 Requires -------- atomic-queue-devel (rpmlib, GLIBC filtered): Provides -------- atomic-queue-devel: atomic-queue-devel atomic-queue-devel(x86-64) atomic-queue-static Generated by fedora-review 0.7.6 (b083f91) last change: 2020-11-10 Command line :/usr/bin/fedora-review -b 1988151 -m fedora-rawhide-x86_64 Buildroot used: fedora-rawhide-x86_64 Active plugins: C/C++, Generic, Shell-api Disabled plugins: Python, Perl, fonts, PHP, Java, Haskell, Ruby, SugarActivity, Ocaml, R Disabled flags: EPEL6, EPEL7, DISTTAG, BATCH, EXARCH
Thanks for the review! > - Note the description-line-too-long warning from rpmlint. There is a macro on > the indicated line that expands to a fairly long string. Thanks! I’ll fix this. > - Regarding support for ppc64le and s390x, it looks like the only assembly is > in defs.h, used to define the spin_loop_pause() function, right? Correct, this is the only bit that has to be written to enable additional architectures. I’d love to be able to fill in the missing architectures. > If that is > the case, then it would be correct (although unfriendly to the CPU) to define > an empty spin_loop_pause() function, or to expand to whatever the no-op > instruction is for each platform. For PowerPC, something like this (largely > stolen from the Linux kernel's arch/powerpc/include/asm/vdso/processor.h): > > […] > > I know you aren't upstream, just saying that if you want to see the > ExclusiveArch go away, it may not be too difficult to make that happen. Your suggestions are probably correct, but since this package is MIT-licensed, I can’t use anything copied from the GPL-licensed Linux kernel, and I don’t know either missing architecture well enough to craft something I am confident in from first principles.
(fedscm-admin): The Pagure repository was created at https://src.fedoraproject.org/rpms/atomic-queue
(In reply to Ben Beasley from comment #4) > Your suggestions are probably correct, but since this package is > MIT-licensed, I can’t use anything copied from the GPL-licensed Linux > kernel, and I don’t know either missing architecture well enough to craft > something I am confident in from first principles. Fair enough. If the missing architectures ever become burdensome, though, simply supplying an empty spin_loop_pause() function should be correct, although you may be able to cook dinner on your CPU if you actually run the code. :-)
I’ll go ahead and create the tracker bugs for the missing architectures. It might still be possible to resolve them with some care, and perhaps consultation with upstream. I think I now basically understand the requirements for a minimal “empty” spin_loop_pause(). The fallback for ARMs that don’t support the “yield” instruction is a good example. It’s a “nop” instruction plus a memory clobber, to create a “compiler barrier” that keeps the compiler from eliding, hoisting, or otherwise subverting the pause function. This is exactly what the example you gave for PowerPC is doing. “or 1 1 1” is an idiomatic nop for PowerPC, or’ing a register with itself to waste time with no effect. I’m not so sure about the s390x example from Linux. As far as I can understand from the limited documentation I’ve found, it’s a conditional branch where the condition is zero so the branch is not taken. However, it seems the first argument somehow causes the BCR instruction to have a synchronization function as well (https://github.com/golang/go/issues/42479), although I haven’t found good documentation on the exact effects. I assume this is intentional, and I wonder why it is needed.
FEDORA-2021-4b83d81872 has been submitted as an update to Fedora 34. https://bodhi.fedoraproject.org/updates/FEDORA-2021-4b83d81872
FEDORA-2021-69a5c768aa has been submitted as an update to Fedora 33. https://bodhi.fedoraproject.org/updates/FEDORA-2021-69a5c768aa
FEDORA-EPEL-2021-076a96c8a6 has been submitted as an update to Fedora EPEL 8. https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2021-076a96c8a6
FEDORA-2021-69a5c768aa has been pushed to the Fedora 33 testing repository. Soon you'll be able to install the update with the following command: `sudo dnf install --enablerepo=updates-testing --advisory=FEDORA-2021-69a5c768aa \*` You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2021-69a5c768aa See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates.
FEDORA-EPEL-2021-076a96c8a6 has been pushed to the Fedora EPEL 8 testing repository. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2021-076a96c8a6 See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates.
FEDORA-2021-4b83d81872 has been pushed to the Fedora 34 testing repository. Soon you'll be able to install the update with the following command: `sudo dnf install --enablerepo=updates-testing --advisory=FEDORA-2021-4b83d81872 \*` You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2021-4b83d81872 See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates.
FEDORA-EPEL-2021-076a96c8a6 has been pushed to the Fedora EPEL 8 stable repository. If problem still persists, please make note of it in this bug report.
FEDORA-2021-4b83d81872 has been pushed to the Fedora 34 stable repository. If problem still persists, please make note of it in this bug report.
FEDORA-2021-69a5c768aa has been pushed to the Fedora 33 stable repository. If problem still persists, please make note of it in this bug report.