tests/common: Remove urcu-signal common test files
[urcu.git] / README.md
... / ...
CommitLineData
1<!--
2SPDX-FileCopyrightText: 2023 EfficiOS Inc.
3
4SPDX-License-Identifier: CC-BY-4.0
5-->
6
7Userspace RCU Implementation
8============================
9
10by Mathieu Desnoyers and Paul E. McKenney
11
12
13Building
14--------
15
16 ./bootstrap # skip if using tarball
17 ./configure
18 make
19 make install
20 ldconfig
21
22Hints:
23
24 - Forcing 32-bit build:
25
26 CFLAGS="-m32 -g -O2" ./configure
27
28 - Forcing 64-bit build:
29
30 CFLAGS="-m64 -g -O2" ./configure
31
32 - Forcing a 32-bit build with 386 backward compatibility:
33
34 CFLAGS="-m32 -g -O2" ./configure --host=i386-pc-linux-gnu
35
36 - Forcing a 32-bit build for Sparcv9 (typical for Sparc v9)
37
38 CFLAGS="-m32 -Wa,-Av9a -g -O2" ./configure
39
40
41Architectures supported
42-----------------------
43
44Currently, the following architectures are supported:
45
46 - x86 (i386, i486, i586, i686)
47 - amd64 / x86\_64
48 - PowerPC 32/64
49 - S390, S390x
50 - ARM 32/64
51 - MIPS
52 - NIOS2
53 - Alpha
54 - ia64
55 - Sparcv9 32/64
56 - Tilera
57 - hppa/PA-RISC
58 - m68k
59 - RISC-V
60
61Tested on:
62
63 - Linux all architectures
64 - FreeBSD 13 i386/amd64
65 - Cygwin i386/amd64
66 - MacOS amd64/arm64
67
68Should also work on:
69
70 - Android
71 - NetBSD 5
72 - OpenBSD
73 - Solaris
74
75(more testing needed before claiming support for these OS).
76
77
78Toolchain support
79-----------------
80
81The C compiler used needs to support at least C99. The C++ compiler used needs
82to support at least C++11. The oldest GCC version officialy supported and
83tested is 4.8.
84
85Older GCC versions might still work with the following exceptions:
86
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.
100
101Clang version 3.0 (based on LLVM 3.0) is supported.
102
103Glibc >= 2.4 should work but the older version we test against is
104currently 2.17.
105
106
107Build system
108------------
109
110For developers using the Git tree:
111
112This source tree is based on the autotools suite from GNU to simplify
113portability. Here are some things you should have on your system in order to
114compile the git repository tree :
115
116 - GNU autotools (automake >=1.12, autoconf >=2.69)
117 (make sure your system wide `automake` points to a recent version!)
118 - GNU Libtool >=2.2
119 (for more information, go to <http://www.gnu.org/software/autoconf/>)
120
121If you get the tree from the repository, you will need to use the `bootstrap`
122script in the root of the tree. It calls all the GNU tools needed to prepare
123the tree configuration.
124
125Test scripts provided in the `tests/` directory of the source tree depend
126on `bash` and the `seq` program.
127
128
129API
130---
131
132See the relevant API documentation files in `doc/`. The APIs provided by
133Userspace RCU are, by prefix:
134
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))
142
143
144Quick start guide
145-----------------
146
147### Usage of all urcu libraries:
148
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.
160
161Those small inlined functions are guaranteed to match the library
162content as long as the library major version is unchanged.
163Therefore, the application *must* be compiled with headers matching
164the library major version number. Applications using
165`URCU_INLINE_SMALL_FUNCTIONS` may be unable to use debugging
166features of Userspace RCU without being recompiled.
167
168There are multiple flavors of liburcu available:
169
170 - `memb`,
171 - `qsbr`,
172 - `mb`,
173 - `signal`,
174 - `bp`.
175
176The API members start with the prefix `urcu_<flavor>_`, where
177`<flavor>` is the chosen flavor name.
178
179
180### Usage of `liburcu-memb`
181
182 1. `#include <urcu/urcu-memb.h>`
183 2. Link the application with `-lurcu-memb`
184
185This is the preferred version of the library, in terms of
186grace-period detection speed, read-side speed and flexibility.
187Dynamically detects kernel support for `sys_membarrier()`. Falls back
188on `urcu-mb` scheme if support is not present, which has slower
189read-side. Use the `--disable-sys-membarrier-fallback` configure option
190to disable the fall back, thus requiring `sys_membarrier()` to be
191available. This gives a small speedup when `sys_membarrier()` is
192supported by the kernel, and aborts in the library constructor if not
193supported.
194
195
196### Usage of `liburcu-qsbr`
197
198 1. `#include <urcu/urcu-qsbr.h>`
199 2. Link with `-lurcu-qsbr`
200
201The QSBR flavor of RCU needs to have each reader thread executing
202`rcu_quiescent_state()` periodically to progress. `rcu_thread_online()`
203and `rcu_thread_offline()` can be used to mark long periods for which
204the threads are not active. It provides the fastest read-side at the
205expense of more intrusiveness in the application code.
206
207
208### Usage of `liburcu-mb`
209
210 1. `#include <urcu/urcu-mb.h>`
211 2. Link with `-lurcu-mb`
212
213This version of the urcu library uses memory barriers on the writer
214and reader sides. This results in faster grace-period detection, but
215results in slower reads.
216
217
218### Usage of `liburcu-signal`
219
220 1. `#include <urcu/urcu-signal.h>`
221 2. Link the application with `-lurcu-signal`
222
223NOTE: The `liburcu-signal` flavor is *deprecated* and will be removed in the
224future. It is now identical to `liburcu-mb` at the exception of the symbols and
225public header files. It is therefore slower than previous versions. Users are
226encouraged to migrate to the `liburcu-memb` flavor.
227
228### Usage of `liburcu-bp`
229
230 1. `#include <urcu/urcu-bp.h>`
231 2. Link with `-lurcu-bp`
232
233The BP library flavor stands for "bulletproof". It is specifically
234designed to help tracing library to hook on applications without
235requiring to modify these applications. `urcu_bp_init()`, and
236`urcu_bp_unregister_thread()` all become nops, whereas calling
237`urcu_bp_register_thread()` becomes optional. The state is dealt with by
238the library internally at the expense of read-side and write-side
239performance.
240
241
242### Initialization
243
244Each thread that has reader critical sections (that uses
245`urcu_<flavor>_read_lock()`/`urcu_<flavor>_read_unlock()` must first
246register to the URCU library. This is done by calling
247`urcu_<flavor>_register_thread()`. Unregistration must be performed
248before exiting the thread by using `urcu_<flavor>_unregister_thread()`.
249
250
251### Reading
252
253Reader critical sections must be protected by locating them between
254calls to `urcu_<flavor>_read_lock()` and `urcu_<flavor>_read_unlock()`.
255Inside that lock, `rcu_dereference()` may be called to read an RCU
256protected pointer.
257
258
259### Writing
260
261`rcu_assign_pointer()` and `rcu_xchg_pointer()` may be called anywhere.
262After, `urcu_<flavor>_synchronize_rcu()` must be called. When it
263returns, the old values are not in usage anymore.
264
265As an alternative to `urcu_<flavor>_synchronize_rcu()`,
266it is also possible to use the urcu polling mechanism to wait for a
267grace period to elapse. This can be done by using
268`urcu_<flavor>_start_poll_synchronize_rcu()`
269to start the grace period polling, and then invoke
270`urcu_<flavor>_poll_state_synchronize_rcu()`, which returns true if
271the grace period has completed, false otherwise.
272
273
274### Usage of `liburcu-defer`
275
276 - Follow instructions for either `liburcu-memb`, `liburcu-qsbr`,
277 `liburcu-mb`, `liburcu-signal`, or `liburcu-bp` above.
278 The `liburcu-defer` functionality is pulled into each of
279 those library modules.
280 - Provides `urcu_<flavor>_defer_rcu()` primitive to enqueue delayed
281 callbacks. Queued callbacks are executed in batch periodically after
282 a grace period. Do _not_ use `urcu_<flavor>_defer_rcu()` within a
283 read-side critical section, because it may call
284 `urcu_<flavor>_synchronize_rcu()` if the thread queue is full. This
285 can lead to deadlock or worse.
286 - Requires that `urcu_<flavor>_defer_barrier()` must be called in
287 library destructor if a library queues callbacks and is expected to
288 be unloaded with `dlclose()`.
289
290Its API is currently experimental. It may change in future library releases.
291
292
293### Usage of `urcu-call-rcu`
294
295 - Follow instructions for either `liburcu-memb`, `liburcu-qsbr`,
296 `liburcu-mb`, `liburcu-signal`, or `liburcu-bp` above.
297 The `urcu-call-rcu` functionality is pulled into each of
298 those library modules.
299 - Provides the `urcu_<flavor>_call_rcu()` primitive to enqueue delayed
300 callbacks in a manner similar to `urcu_<flavor>_defer_rcu()`, but
301 without ever delaying for a grace period. On the other hand,
302 `urcu_<flavor>_call_rcu()`'s best-case overhead is not quite as good
303 as that of `urcu_<flavor>_defer_rcu()`.
304 - Provides `urcu_<flavor>_call_rcu()` to allow asynchronous handling
305 of RCU grace periods. A number of additional functions are provided
306 to manage the helper threads used by `urcu_<flavor>_call_rcu()`, but
307 reasonable defaults are used if these additional functions are not
308 invoked. See [`doc/rcu-api.md`](doc/rcu-api.md) in userspace-rcu
309 documentation for more details.
310
311
312### Being careful with signals
313
314The `liburcu-signal` library uses signals internally. The signal handler is
315registered with the `SA_RESTART` flag. However, these signals may cause
316some non-restartable system calls to fail with `errno = EINTR`. Care
317should be taken to restart system calls manually if they fail with this
318error. A list of non-restartable system calls may be found in
319`signal(7)`.
320
321Read-side critical sections are allowed in a signal handler,
322except those setup with `sigaltstack(2)`, with `liburcu-memb` and
323`liburcu-mb`. Be careful, however, to disable these signals
324between thread creation and calls to `urcu_<flavor>_register_thread()`,
325because a signal handler nesting on an unregistered thread would not be
326allowed to call `urcu_<flavor>_read_lock()`.
327
328Read-side critical sections are _not_ allowed in a signal handler with
329`liburcu-qsbr`, unless signals are disabled explicitly around each
330`urcu_qsbr_quiescent_state()` calls, when threads are put offline and around
331calls to `urcu_qsbr_synchronize_rcu()`. Even then, we do not recommend it.
332
333
334### Interaction with mutexes
335
336One must be careful to do not cause deadlocks due to interaction of
337`urcu_<flavor>_synchronize_rcu()` and RCU read-side with mutexes. If
338`urcu_<flavor>_synchronize_rcu()` is called with a mutex held, this
339mutex (or any mutex which has this mutex in its dependency chain) should
340not be acquired from within a RCU read-side critical section.
341
342This is especially important to understand in the context of the
343QSBR flavor: a registered reader thread being "online" by
344default should be considered as within a RCU read-side critical
345section unless explicitly put "offline". Therefore, if
346`urcu_qsbr_synchronize_rcu()` is called with a mutex held, this mutex,
347as well as any mutex which has this mutex in its dependency chain should
348only be taken when the RCU reader thread is "offline" (this can be
349performed by calling `urcu_qsbr_thread_offline()`).
350
351
352### Interaction with `fork()`
353
354Special care must be taken for applications performing `fork()` without
355any following `exec()`. This is caused by the fact that Linux only clones
356the thread calling `fork()`, and thus never replicates any of the other
357parent thread into the child process. Most `liburcu` implementations
358require that all registrations (as reader, `defer_rcu` and `call_rcu`
359threads) should be released before a `fork()` is performed, except for the
360rather common scenario where `fork()` is immediately followed by `exec()` in
361the child process. The only implementation not subject to that rule is
362`liburcu-bp`, which is designed to handle `fork()` by calling
363`urcu_bp_before_fork`, `urcu_bp_after_fork_parent` and
364`urcu_bp_after_fork_child`.
365
366Applications that use `urcu_<flavor>_call_rcu()` and that `fork()`
367without doing an immediate `exec()` must take special action. The
368parent must invoke `urcu_<flavor>_call_rcu_before_fork()` before the
369`fork()` and `urcu_<flavor>_call_rcu_after_fork_parent()` after the
370`fork()`. The child process must invoke
371`urcu_<flavor>_call_rcu_after_fork_child()`. Even though these three
372APIs are suitable for passing to `pthread_atfork()`, use of
373`pthread_atfork()` is **STRONGLY DISCOURAGED** for programs calling the
374glibc memory allocator (`malloc()`, `calloc()`, `free()`, ...) within
375`urcu_<flavor>_call_rcu` callbacks. This is due to limitations in the
376way glibc memory allocator handles calls to the memory allocator from
377concurrent threads while the `pthread_atfork()` handlers are executing.
378
379Combining e.g.:
380
381 - call to `free()` from callbacks executed within
382 `urcu_<flavor>_call_rcu` worker threads,
383 - executing `urcu_<flavor>_call_rcu` atfork handlers within the glibc
384 pthread atfork mechanism,
385
386will sometimes trigger interesting process hangs. This usually
387hangs on a memory allocator lock within glibc.
388
389
390### Thread Local Storage (TLS)
391
392Userspace RCU can fall back on `pthread_getspecific()` to emulate
393TLS variables on systems where it is not available. This behavior
394can be forced by specifying `--disable-compiler-tls` as configure
395argument.
396
397
398### Usage of `DEBUG_RCU` & `--enable-rcu-debug`
399
400By default the library is configured with internal debugging
401self-checks disabled.
402
403For always-on debugging self-checks:
404
405 ./configure --enable-rcu-debug
406
407For fine grained enabling of debugging self-checks, build
408userspace-rcu with `DEBUG_RCU` defined and compile dependent
409applications with `DEBUG_RCU` defined when necessary.
410
411Warning: Enabling this feature result in a performance penalty.
412
413
414### Usage of `DEBUG_YIELD`
415
416`DEBUG_YIELD` is used to add random delays in the code for testing
417purposes.
418
419
420### SMP support
421
422By default the library is configured to use synchronization primitives
423adequate for SMP systems. On uniprocessor systems, support for SMP
424systems can be disabled with:
425
426 ./configure --disable-smp-support
427
428theoretically yielding slightly better performance.
429
430
431### Usage of `--enable-cds-lfht-iter-debug`
432
433By default the library is configured with extra debugging checks for
434lock-free hash table iterator traversal disabled.
435
436Building liburcu with `--enable-cds-lfht-iter-debug` and rebuilding
437application to match the ABI change allows finding cases where the hash
438table iterator is re-purposed to be used on a different hash table while
439still being used to iterate on a hash table.
440
441This option alters the rculfhash ABI. Make sure to compile both library
442and application with matching configuration.
443
444### Usage of `--enable-compiler-atomic-builtins`
445
446Building liburcu with `--enable-compiler-atomic-builtins` implements the uatomic
447API with the compiler atomic builtins if supported.
448
449Make targets
450------------
451
452In addition to the usual `make check` target, Userspace RCU features
453`make regtest`, `make short_bench` and `make long_bench` targets:
454
455 - `make check`: short tests, meant to be run when rebuilding or
456 porting Userspace RCU.
457 - `make regtest`: long (many hours) test, meant to be run when
458 modifying Userspace RCU or porting it to a new architecture or
459 operating system.
460 - `make short_bench`: short benchmarks, 3 seconds per test.
461 - `make long_bench`: long (many hours) benchmarks, 30 seconds per test.
462
463
464Known issues
465------------
466
467There is an application vs library compatibility issue between
468applications built using Userspace RCU 0.10 headers linked against
469Userspace RCU 0.11 or 0.12 shared objects. The problem occurs as
470follows:
471
472 - An application executable is built with `_LGPL_SOURCE` defined, includes
473 any of the Userspace RCU 0.10 urcu flavor headers, and is built
474 without the `-fpic` compiler option.
475
476 - The Userspace RCU 0.10 library shared objects are updated to 0.11
477 or 0.12 without rebuilding the application.
478
479 - The application will hang, typically when RCU grace period
480 (synchronize_rcu) is invoked.
481
482Some possible work-arounds for this are:
483
484 - Rebuild the application against Userspace RCU 0.11+.
485
486 - Rebuild the application with `-fpic`.
487
488 - Upgrade Userspace RCU to 0.13+ without installing 0.11 nor 0.12.
489
490
491Contacts
492--------
493
494You can contact the maintainers on the following mailing list:
495`lttng-dev@lists.lttng.org`.
This page took 0.024855 seconds and 4 git commands to generate.