2 SPDX-FileCopyrightText: 2023 EfficiOS Inc.
4 SPDX-License-Identifier: CC-BY-4.0
7 Userspace RCU Implementation
8 ============================
10 by Mathieu Desnoyers and Paul E. McKenney
16 ./bootstrap # skip if using tarball
24 - Forcing 32-bit build:
26 CFLAGS="-m32 -g -O2" ./configure
28 - Forcing 64-bit build:
30 CFLAGS="-m64 -g -O2" ./configure
32 - Forcing a 32-bit build with 386 backward compatibility:
34 CFLAGS="-m32 -g -O2" ./configure --host=i386-pc-linux-gnu
36 - Forcing a 32-bit build for Sparcv9 (typical for Sparc v9)
38 CFLAGS="-m32 -Wa,-Av9a -g -O2" ./configure
41 Architectures supported
42 -----------------------
44 Currently, the following architectures are supported:
46 - x86 (i386, i486, i586, i686)
63 - Linux all architectures
64 - FreeBSD 13 i386/amd64
75 (more testing needed before claiming support for these OS).
81 The C compiler used needs to support at least C99. The C++ compiler used needs
82 to support at least C++11. The oldest GCC version officialy supported and
85 Older GCC versions might still work with the following exceptions:
87 - GCC 3.3 and 3.4 have a bug that prevents them from generating volatile
88 accesses to offsets in a TLS structure on 32-bit x86. These versions are
89 therefore not compatible with `liburcu` on x86 32-bit
90 (i386, i486, i586, i686).
91 The problem has been reported to the GCC community:
92 <http://www.mail-archive.com/gcc-bugs@gcc.gnu.org/msg281255.html>
93 - GCC 3.3 cannot match the "xchg" instruction on 32-bit x86 build.
94 See <http://kerneltrap.org/node/7507>
95 - Alpha, ia64 and ARM architectures depend on GCC 4.x with atomic builtins
96 support. For ARM this was introduced with GCC 4.4:
97 <http://gcc.gnu.org/gcc-4.4/changes.html>.
98 - Linux aarch64 depends on GCC 5.1 or better because prior versions
99 perform unsafe access to deallocated stack.
101 Clang version 3.0 (based on LLVM 3.0) is supported.
103 Glibc >= 2.4 should work but the older version we test against is
110 For developers using the Git tree:
112 This source tree is based on the autotools suite from GNU to simplify
113 portability. Here are some things you should have on your system in order to
114 compile the git repository tree :
116 - GNU autotools (automake >=1.12, autoconf >=2.69)
117 (make sure your system wide `automake` points to a recent version!)
119 (for more information, go to <http://www.gnu.org/software/autoconf/>)
121 If you get the tree from the repository, you will need to use the `bootstrap`
122 script in the root of the tree. It calls all the GNU tools needed to prepare
123 the tree configuration.
125 Test scripts provided in the `tests/` directory of the source tree depend
126 on `bash` and the `seq` program.
132 See the relevant API documentation files in `doc/`. The APIs provided by
133 Userspace RCU are, by prefix:
135 - `rcu_`: Read-Copy Update (see [`doc/rcu-api.md`](doc/rcu-api.md))
136 - `cmm_`: Concurrent Memory Model
137 - `caa_`: Concurrent Architecture Abstraction
138 - `cds_`: Concurrent Data Structures
139 (see [`doc/cds-api.md`](doc/cds-api.md))
140 - `uatomic_`: Userspace Atomic
141 (see [`doc/uatomic-api.md`](doc/uatomic-api.md))
147 ### Usage of all urcu libraries:
149 - Define `_LGPL_SOURCE` (only) if your code is LGPL or GPL compatible
150 before including the `urcu.h` or `urcu-qsbr.h` header. If your application
151 is distributed under another license, function calls will be generated
152 instead of inlines, so your application can link with the library.
153 - Linking with one of the libraries below is always necessary even for
154 LGPL and GPL applications.
155 - Define `URCU_INLINE_SMALL_FUNCTIONS` before including Userspace RCU
156 headers if you want Userspace RCU to inline small functions (10
157 lines or less) into the application. It can be used by applications
158 distributed under any kind of license, and does *not* make the
159 application a derived work of Userspace RCU.
161 Those small inlined functions are guaranteed to match the library
162 content as long as the library major version is unchanged.
163 Therefore, the application *must* be compiled with headers matching
164 the library major version number. Applications using
165 `URCU_INLINE_SMALL_FUNCTIONS` may be unable to use debugging
166 features of Userspace RCU without being recompiled.
168 There are multiple flavors of liburcu available:
176 The API members start with the prefix `urcu_<flavor>_`, where
177 `<flavor>` is the chosen flavor name.
180 ### Usage of `liburcu-memb`
182 1. `#include <urcu/urcu-memb.h>`
183 2. Link the application with `-lurcu-memb`
185 This is the preferred version of the library, in terms of
186 grace-period detection speed, read-side speed and flexibility.
187 Dynamically detects kernel support for `sys_membarrier()`. Falls back
188 on `urcu-mb` scheme if support is not present, which has slower
189 read-side. Use the `--disable-sys-membarrier-fallback` configure option
190 to disable the fall back, thus requiring `sys_membarrier()` to be
191 available. This gives a small speedup when `sys_membarrier()` is
192 supported by the kernel, and aborts in the library constructor if not
196 ### Usage of `liburcu-qsbr`
198 1. `#include <urcu/urcu-qsbr.h>`
199 2. Link with `-lurcu-qsbr`
201 The QSBR flavor of RCU needs to have each reader thread executing
202 `rcu_quiescent_state()` periodically to progress. `rcu_thread_online()`
203 and `rcu_thread_offline()` can be used to mark long periods for which
204 the threads are not active. It provides the fastest read-side at the
205 expense of more intrusiveness in the application code.
208 ### Usage of `liburcu-mb`
210 1. `#include <urcu/urcu-mb.h>`
211 2. Link with `-lurcu-mb`
213 This version of the urcu library uses memory barriers on the writer
214 and reader sides. This results in faster grace-period detection, but
215 results in slower reads.
218 ### Usage of `liburcu-signal`
220 1. `#include <urcu/urcu-signal.h>`
221 2. Link the application with `-lurcu-signal`
223 Version of the library that requires a signal, typically `SIGUSR1`. Can
224 be overridden with `-DSIGRCU` by modifying `Makefile.build.inc`.
227 ### Usage of `liburcu-bp`
229 1. `#include <urcu/urcu-bp.h>`
230 2. Link with `-lurcu-bp`
232 The BP library flavor stands for "bulletproof". It is specifically
233 designed to help tracing library to hook on applications without
234 requiring to modify these applications. `urcu_bp_init()`, and
235 `urcu_bp_unregister_thread()` all become nops, whereas calling
236 `urcu_bp_register_thread()` becomes optional. The state is dealt with by
237 the library internally at the expense of read-side and write-side
243 Each thread that has reader critical sections (that uses
244 `urcu_<flavor>_read_lock()`/`urcu_<flavor>_read_unlock()` must first
245 register to the URCU library. This is done by calling
246 `urcu_<flavor>_register_thread()`. Unregistration must be performed
247 before exiting the thread by using `urcu_<flavor>_unregister_thread()`.
252 Reader critical sections must be protected by locating them between
253 calls to `urcu_<flavor>_read_lock()` and `urcu_<flavor>_read_unlock()`.
254 Inside that lock, `rcu_dereference()` may be called to read an RCU
260 `rcu_assign_pointer()` and `rcu_xchg_pointer()` may be called anywhere.
261 After, `urcu_<flavor>_synchronize_rcu()` must be called. When it
262 returns, the old values are not in usage anymore.
264 As an alternative to `urcu_<flavor>_synchronize_rcu()`,
265 it is also possible to use the urcu polling mechanism to wait for a
266 grace period to elapse. This can be done by using
267 `urcu_<flavor>_start_poll_synchronize_rcu()`
268 to start the grace period polling, and then invoke
269 `urcu_<flavor>_poll_state_synchronize_rcu()`, which returns true if
270 the grace period has completed, false otherwise.
273 ### Usage of `liburcu-defer`
275 - Follow instructions for either `liburcu-memb`, `liburcu-qsbr`,
276 `liburcu-mb`, `liburcu-signal`, or `liburcu-bp` above.
277 The `liburcu-defer` functionality is pulled into each of
278 those library modules.
279 - Provides `urcu_<flavor>_defer_rcu()` primitive to enqueue delayed
280 callbacks. Queued callbacks are executed in batch periodically after
281 a grace period. Do _not_ use `urcu_<flavor>_defer_rcu()` within a
282 read-side critical section, because it may call
283 `urcu_<flavor>_synchronize_rcu()` if the thread queue is full. This
284 can lead to deadlock or worse.
285 - Requires that `urcu_<flavor>_defer_barrier()` must be called in
286 library destructor if a library queues callbacks and is expected to
287 be unloaded with `dlclose()`.
289 Its API is currently experimental. It may change in future library releases.
292 ### Usage of `urcu-call-rcu`
294 - Follow instructions for either `liburcu-memb`, `liburcu-qsbr`,
295 `liburcu-mb`, `liburcu-signal`, or `liburcu-bp` above.
296 The `urcu-call-rcu` functionality is pulled into each of
297 those library modules.
298 - Provides the `urcu_<flavor>_call_rcu()` primitive to enqueue delayed
299 callbacks in a manner similar to `urcu_<flavor>_defer_rcu()`, but
300 without ever delaying for a grace period. On the other hand,
301 `urcu_<flavor>_call_rcu()`'s best-case overhead is not quite as good
302 as that of `urcu_<flavor>_defer_rcu()`.
303 - Provides `urcu_<flavor>_call_rcu()` to allow asynchronous handling
304 of RCU grace periods. A number of additional functions are provided
305 to manage the helper threads used by `urcu_<flavor>_call_rcu()`, but
306 reasonable defaults are used if these additional functions are not
307 invoked. See [`doc/rcu-api.md`](doc/rcu-api.md) in userspace-rcu
308 documentation for more details.
311 ### Being careful with signals
313 The `liburcu-signal` library uses signals internally. The signal handler is
314 registered with the `SA_RESTART` flag. However, these signals may cause
315 some non-restartable system calls to fail with `errno = EINTR`. Care
316 should be taken to restart system calls manually if they fail with this
317 error. A list of non-restartable system calls may be found in
320 Read-side critical sections are allowed in a signal handler,
321 except those setup with `sigaltstack(2)`, with `liburcu-memb` and
322 `liburcu-mb`. Be careful, however, to disable these signals
323 between thread creation and calls to `urcu_<flavor>_register_thread()`,
324 because a signal handler nesting on an unregistered thread would not be
325 allowed to call `urcu_<flavor>_read_lock()`.
327 Read-side critical sections are _not_ allowed in a signal handler with
328 `liburcu-qsbr`, unless signals are disabled explicitly around each
329 `urcu_qsbr_quiescent_state()` calls, when threads are put offline and around
330 calls to `urcu_qsbr_synchronize_rcu()`. Even then, we do not recommend it.
333 ### Interaction with mutexes
335 One must be careful to do not cause deadlocks due to interaction of
336 `urcu_<flavor>_synchronize_rcu()` and RCU read-side with mutexes. If
337 `urcu_<flavor>_synchronize_rcu()` is called with a mutex held, this
338 mutex (or any mutex which has this mutex in its dependency chain) should
339 not be acquired from within a RCU read-side critical section.
341 This is especially important to understand in the context of the
342 QSBR flavor: a registered reader thread being "online" by
343 default should be considered as within a RCU read-side critical
344 section unless explicitly put "offline". Therefore, if
345 `urcu_qsbr_synchronize_rcu()` is called with a mutex held, this mutex,
346 as well as any mutex which has this mutex in its dependency chain should
347 only be taken when the RCU reader thread is "offline" (this can be
348 performed by calling `urcu_qsbr_thread_offline()`).
351 ### Interaction with `fork()`
353 Special care must be taken for applications performing `fork()` without
354 any following `exec()`. This is caused by the fact that Linux only clones
355 the thread calling `fork()`, and thus never replicates any of the other
356 parent thread into the child process. Most `liburcu` implementations
357 require that all registrations (as reader, `defer_rcu` and `call_rcu`
358 threads) should be released before a `fork()` is performed, except for the
359 rather common scenario where `fork()` is immediately followed by `exec()` in
360 the child process. The only implementation not subject to that rule is
361 `liburcu-bp`, which is designed to handle `fork()` by calling
362 `urcu_bp_before_fork`, `urcu_bp_after_fork_parent` and
363 `urcu_bp_after_fork_child`.
365 Applications that use `urcu_<flavor>_call_rcu()` and that `fork()`
366 without doing an immediate `exec()` must take special action. The
367 parent must invoke `urcu_<flavor>_call_rcu_before_fork()` before the
368 `fork()` and `urcu_<flavor>_call_rcu_after_fork_parent()` after the
369 `fork()`. The child process must invoke
370 `urcu_<flavor>_call_rcu_after_fork_child()`. Even though these three
371 APIs are suitable for passing to `pthread_atfork()`, use of
372 `pthread_atfork()` is **STRONGLY DISCOURAGED** for programs calling the
373 glibc memory allocator (`malloc()`, `calloc()`, `free()`, ...) within
374 `urcu_<flavor>_call_rcu` callbacks. This is due to limitations in the
375 way glibc memory allocator handles calls to the memory allocator from
376 concurrent threads while the `pthread_atfork()` handlers are executing.
380 - call to `free()` from callbacks executed within
381 `urcu_<flavor>_call_rcu` worker threads,
382 - executing `urcu_<flavor>_call_rcu` atfork handlers within the glibc
383 pthread atfork mechanism,
385 will sometimes trigger interesting process hangs. This usually
386 hangs on a memory allocator lock within glibc.
389 ### Thread Local Storage (TLS)
391 Userspace RCU can fall back on `pthread_getspecific()` to emulate
392 TLS variables on systems where it is not available. This behavior
393 can be forced by specifying `--disable-compiler-tls` as configure
397 ### Usage of `DEBUG_RCU` & `--enable-rcu-debug`
399 By default the library is configured with internal debugging
400 self-checks disabled.
402 For always-on debugging self-checks:
404 ./configure --enable-rcu-debug
406 For fine grained enabling of debugging self-checks, build
407 userspace-rcu with `DEBUG_RCU` defined and compile dependent
408 applications with `DEBUG_RCU` defined when necessary.
410 Warning: Enabling this feature result in a performance penalty.
413 ### Usage of `DEBUG_YIELD`
415 `DEBUG_YIELD` is used to add random delays in the code for testing
421 By default the library is configured to use synchronization primitives
422 adequate for SMP systems. On uniprocessor systems, support for SMP
423 systems can be disabled with:
425 ./configure --disable-smp-support
427 theoretically yielding slightly better performance.
430 ### Usage of `--enable-cds-lfht-iter-debug`
432 By default the library is configured with extra debugging checks for
433 lock-free hash table iterator traversal disabled.
435 Building liburcu with `--enable-cds-lfht-iter-debug` and rebuilding
436 application to match the ABI change allows finding cases where the hash
437 table iterator is re-purposed to be used on a different hash table while
438 still being used to iterate on a hash table.
440 This option alters the rculfhash ABI. Make sure to compile both library
441 and application with matching configuration.
443 ### Usage of `--enable-compiler-atomic-builtins`
445 Building liburcu with `--enable-compiler-atomic-builtins` implements the uatomic
446 API with the compiler atomic builtins if supported.
451 In addition to the usual `make check` target, Userspace RCU features
452 `make regtest`, `make short_bench` and `make long_bench` targets:
454 - `make check`: short tests, meant to be run when rebuilding or
455 porting Userspace RCU.
456 - `make regtest`: long (many hours) test, meant to be run when
457 modifying Userspace RCU or porting it to a new architecture or
459 - `make short_bench`: short benchmarks, 3 seconds per test.
460 - `make long_bench`: long (many hours) benchmarks, 30 seconds per test.
466 There is an application vs library compatibility issue between
467 applications built using Userspace RCU 0.10 headers linked against
468 Userspace RCU 0.11 or 0.12 shared objects. The problem occurs as
471 - An application executable is built with `_LGPL_SOURCE` defined, includes
472 any of the Userspace RCU 0.10 urcu flavor headers, and is built
473 without the `-fpic` compiler option.
475 - The Userspace RCU 0.10 library shared objects are updated to 0.11
476 or 0.12 without rebuilding the application.
478 - The application will hang, typically when RCU grace period
479 (synchronize_rcu) is invoked.
481 Some possible work-arounds for this are:
483 - Rebuild the application against Userspace RCU 0.11+.
485 - Rebuild the application with `-fpic`.
487 - Upgrade Userspace RCU to 0.13+ without installing 0.11 nor 0.12.
493 You can contact the maintainers on the following mailing list:
494 `lttng-dev@lists.lttng.org`.