aarch64: blacklist gcc prior to 5.1
[urcu.git] / README.md
CommitLineData
d589a916
PP
1Userspace RCU Implementation
2============================
3
4by Mathieu Desnoyers and Paul E. McKenney
5
6
7Building
8--------
9
10 ./bootstrap # skip if using tarball
11 ./configure
12 make
13 make install
14 ldconfig
15
16Hints:
17
18 - Forcing 32-bit build:
19
20 CFLAGS="-m32 -g -O2" ./configure
21
22 - Forcing 64-bit build:
23
24 CFLAGS="-m64 -g -O2" ./configure
25
26 - Forcing a 32-bit build with 386 backward compatibility:
27
28 CFLAGS="-m32 -g -O2" ./configure --host=i386-pc-linux-gnu
29
30 - Forcing a 32-bit build for Sparcv9 (typical for Sparc v9)
31
32 CFLAGS="-m32 -Wa,-Av9a -g -O2" ./configure
33
34
35Architectures supported
36-----------------------
37
38Currently, the following architectures are supported:
39
f328865f
MD
40 - x86 (i386, i486, i586, i686)
41 - amd64 / x86_64
d589a916
PP
42 - PowerPC 32/64
43 - S390, S390x
44 - ARM 32/64
45 - MIPS
859050b3 46 - NIOS2
d589a916
PP
47 - Alpha
48 - ia64
49 - Sparcv9 32/64
50 - Tilera
51 - hppa/PA-RISC
f328865f
MD
52 - m68k
53 - RISC-V
54
55Tested on:
56
57 - Linux all architectures
58 - FreeBSD 8.2/8.3/9.0/9.1/10.0 i386/amd64
59 - Solaris 10/11 i386
60 - Cygwin i386/amd64
61 - MacOSX amd64
d589a916 62
d589a916
PP
63Should also work on:
64
65 - Android
66 - NetBSD 5
67 - OpenBSD
68 - Darwin
69
70(more testing needed before claiming support for these OS).
71
72Linux ARM depends on running a Linux kernel 2.6.15 or better, GCC 4.4 or
73better.
74
75The GCC compiler versions 3.3, 3.4, 4.0, 4.1, 4.2, 4.3, 4.4 and 4.5 are
76supported, with the following exceptions:
77
78 - GCC 3.3 and 3.4 have a bug that prevents them from generating volatile
79 accesses to offsets in a TLS structure on 32-bit x86. These versions are
80 therefore not compatible with `liburcu` on x86 32-bit
81 (i386, i486, i586, i686).
82 The problem has been reported to the GCC community:
83 http://www.mail-archive.com/gcc-bugs@gcc.gnu.org/msg281255.html
84 - GCC 3.3 cannot match the "xchg" instruction on 32-bit x86 build.
85 See http://kerneltrap.org/node/7507
86 - Alpha, ia64 and ARM architectures depend on GCC 4.x with atomic builtins
87 support. For ARM this was introduced with GCC 4.4:
88 http://gcc.gnu.org/gcc-4.4/changes.html.
ddec79fd
MD
89 - Linux aarch64 depends on GCC 5.1 or better because prior versions
90 perform unsafe access to deallocated stack.
d589a916
PP
91
92Clang version 3.0 (based on LLVM 3.0) is supported.
93
94Building on MacOS X (Darwin) requires a work-around for processor
95detection:
96
97 - 32-bit:
98
99 ./configure --build=i686-apple-darwin11
100
101 - 64-bit:
102
103 ./configure --build=x86_64-apple-darwin11
104
105For developers using the Git tree:
106
107This source tree is based on the autotools suite from GNU to simplify
108portability. Here are some things you should have on your system in order to
109compile the git repository tree :
110
111 - GNU autotools (automake >=1.10, autoconf >=2.50, autoheader >=2.50)
112 (make sure your system wide `automake` points to a recent version!)
113 - GNU Libtool >=2.2
114 (for more information, go to http://www.gnu.org/software/autoconf/)
115
116If you get the tree from the repository, you will need to use the `bootstrap`
117script in the root of the tree. It calls all the GNU tools needed to prepare
118the tree configuration.
119
120Test scripts provided in the `tests/` directory of the source tree depend
121on `bash` and the `seq` program.
122
123
124API
125---
126
127See the relevant API documentation files in `doc/`. The APIs provided by
128Userspace RCU are, by prefix:
129
dcb9c05a 130 - `rcu_`: Read-Copy Update (see [`doc/rcu-api.md`](doc/rcu-api.md))
d589a916
PP
131 - `cmm_`: Concurrent Memory Model
132 - `caa_`: Concurrent Architecture Abstraction
133 - `cds_`: Concurrent Data Structures
dcb9c05a 134 (see [`doc/cds-api.md`](doc/cds-api.md))
d589a916 135 - `uatomic_`: Userspace Atomic
dcb9c05a 136 (see [`doc/uatomic-api.md`](doc/uatomic-api.md))
d589a916
PP
137
138
139Quick start guide
140-----------------
141
142### Usage of all urcu libraries:
143
144 - Define `_LGPL_SOURCE` (only) if your code is LGPL or GPL compatible
145 before including the `urcu.h` or `urcu-qsbr.h` header. If your application
146 is distributed under another license, function calls will be generated
147 instead of inlines, so your application can link with the library.
148 - Linking with one of the libraries below is always necessary even for
149 LGPL and GPL applications.
150 - Define `URCU_INLINE_SMALL_FUNCTIONS` before including Userspace RCU
151 headers if you want Userspace RCU to inline small functions (10
152 lines or less) into the application. It can be used by applications
153 distributed under any kind of license, and does *not* make the
154 application a derived work of Userspace RCU.
155
156Those small inlined functions are guaranteed to match the library
157content as long as the library major version is unchanged.
158Therefore, the application *must* be compiled with headers matching
159the library major version number. Applications using
160`URCU_INLINE_SMALL_FUNCTIONS` may be unable to use debugging
161features of Userspace RCU without being recompiled.
162
f328865f
MD
163There are multiple flavors of liburcu available:
164
165 - `memb`,
166 - `qsbr`,
167 - `mb`,
168 - `signal`,
169 - `bp`.
170
171The API members start with the prefix "urcu_<flavor>_", where
172<flavor> is the chosen flavor name.
173
d589a916 174
f328865f 175### Usage of `liburcu-memb`
d589a916 176
f328865f
MD
177 1. `#include <urcu/urcu-memb.h>`
178 2. Link the application with `-lurcu-memb`
d589a916
PP
179
180This is the preferred version of the library, in terms of
181grace-period detection speed, read-side speed and flexibility.
182Dynamically detects kernel support for `sys_membarrier()`. Falls back
183on `urcu-mb` scheme if support is not present, which has slower
d8d9a340
MD
184read-side. Use the --disable-sys-membarrier-fallback configure option
185to disable the fall back, thus requiring `sys_membarrier()` to be
186available. This gives a small speedup when `sys_membarrier()` is
187supported by the kernel, and aborts in the library constructor if not
188supported.
d589a916
PP
189
190
191### Usage of `liburcu-qsbr`
192
f328865f 193 1. `#include <urcu/urcu-qsbr.h>`
d589a916
PP
194 2. Link with `-lurcu-qsbr`
195
196The QSBR flavor of RCU needs to have each reader thread executing
197`rcu_quiescent_state()` periodically to progress. `rcu_thread_online()`
198and `rcu_thread_offline()` can be used to mark long periods for which
199the threads are not active. It provides the fastest read-side at the
200expense of more intrusiveness in the application code.
201
202
203### Usage of `liburcu-mb`
204
f328865f
MD
205 1. `#include <urcu/urcu-mb.h>`
206 2. Link with `-lurcu-mb`
d589a916
PP
207
208This version of the urcu library uses memory barriers on the writer
209and reader sides. This results in faster grace-period detection, but
210results in slower reads.
211
212
213### Usage of `liburcu-signal`
214
f328865f
MD
215 1. `#include <urcu/urcu-signal.h>`
216 2. Link the application with `-lurcu-signal`
d589a916
PP
217
218Version of the library that requires a signal, typically `SIGUSR1`. Can
219be overridden with `-DSIGRCU` by modifying `Makefile.build.inc`.
220
221
222### Usage of `liburcu-bp`
223
f328865f 224 1. `#include <urcu/urcu-bp.h>`
d589a916
PP
225 2. Link with `-lurcu-bp`
226
227The BP library flavor stands for "bulletproof". It is specifically
228designed to help tracing library to hook on applications without
5b46e39d
MD
229requiring to modify these applications. `urcu_bp_init()`, and
230`urcu_bp_unregister_thread()` all become nops, whereas calling
231`urcu_bp_register_thread()` becomes optional. The state is dealt with by
232the library internally at the expense of read-side and write-side
233performance.
d589a916
PP
234
235
236### Initialization
237
238Each thread that has reader critical sections (that uses
f328865f
MD
239`urcu_<flavor>_read_lock()`/`urcu_<flavor>_read_unlock()` must first
240register to the URCU library. This is done by calling
241`urcu_<flavor>_register_thread()`. Unregistration must be performed
242before exiting the thread by using `urcu_<flavor>_unregister_thread()`.
d589a916
PP
243
244
245### Reading
246
247Reader critical sections must be protected by locating them between
f328865f
MD
248calls to `urcu_<flavor>_read_lock()` and `urcu_<flavor>_read_unlock()`.
249Inside that lock, `rcu_dereference()` may be called to read an RCU
250protected pointer.
d589a916
PP
251
252
253### Writing
254
255`rcu_assign_pointer()` and `rcu_xchg_pointer()` may be called anywhere.
f328865f
MD
256After, `urcu_<flavor>_synchronize_rcu()` must be called. When it
257returns, the old values are not in usage anymore.
d589a916
PP
258
259
260### Usage of `liburcu-defer`
261
f328865f 262 - Follow instructions for either `liburcu-memb`, `liburcu-qsbr`,
d589a916
PP
263 `liburcu-mb`, `liburcu-signal`, or `liburcu-bp` above.
264 The `liburcu-defer` functionality is pulled into each of
265 those library modules.
f328865f
MD
266 - Provides `urcu_<flavor>_defer_rcu()` primitive to enqueue delayed
267 callbacks. Queued callbacks are executed in batch periodically after
268 a grace period. Do _not_ use `urcu_<flavor>_defer_rcu()` within a
269 read-side critical section, because it may call
270 `urcu_<flavor>_synchronize_rcu()` if the thread queue is full. This
271 can lead to deadlock or worse.
272 - Requires that `urcu_<flavor>_defer_barrier()` must be called in
273 library destructor if a library queues callbacks and is expected to
274 be unloaded with `dlclose()`.
d589a916
PP
275
276Its API is currently experimental. It may change in future library releases.
277
278
279### Usage of `urcu-call-rcu`
280
f328865f 281 - Follow instructions for either `liburcu-memb`, `liburcu-qsbr`,
d589a916
PP
282 `liburcu-mb`, `liburcu-signal`, or `liburcu-bp` above.
283 The `urcu-call-rcu` functionality is pulled into each of
284 those library modules.
f328865f
MD
285 - Provides the `urcu_<flavor>_call_rcu()` primitive to enqueue delayed
286 callbacks in a manner similar to `urcu_<flavor>_defer_rcu()`, but
287 without ever delaying for a grace period. On the other hand,
288 `urcu_<flavor>_call_rcu()`'s best-case overhead is not quite as good
289 as that of `urcu_<flavor>_defer_rcu()`.
290 - Provides `urcu_<flavor>_call_rcu()` to allow asynchronous handling
291 of RCU grace periods. A number of additional functions are provided
292 to manage the helper threads used by `urcu_<flavor>_call_rcu()`, but
293 reasonable defaults are used if these additional functions are not
294 invoked. See [`doc/rcu-api.md`](doc/rcu-api.md) in userspace-rcu
295 documentation for more details.
d589a916
PP
296
297
298### Being careful with signals
299
f328865f 300The `liburcu-signal` library uses signals internally. The signal handler is
d589a916
PP
301registered with the `SA_RESTART` flag. However, these signals may cause
302some non-restartable system calls to fail with `errno = EINTR`. Care
303should be taken to restart system calls manually if they fail with this
304error. A list of non-restartable system calls may be found in
f328865f 305`signal(7)`.
d589a916
PP
306
307Read-side critical sections are allowed in a signal handler,
f328865f 308except those setup with `sigaltstack(2)`, with `liburcu-memb` and
d589a916 309`liburcu-mb`. Be careful, however, to disable these signals
f328865f
MD
310between thread creation and calls to `urcu_<flavor>_register_thread()`,
311because a signal handler nesting on an unregistered thread would not be
312allowed to call `urcu_<flavor>_read_lock()`.
d589a916
PP
313
314Read-side critical sections are _not_ allowed in a signal handler with
315`liburcu-qsbr`, unless signals are disabled explicitly around each
f328865f
MD
316`urcu_qsbr_quiescent_state()` calls, when threads are put offline and around
317calls to `urcu_qsbr_synchronize_rcu()`. Even then, we do not recommend it.
d589a916
PP
318
319
320### Interaction with mutexes
321
322One must be careful to do not cause deadlocks due to interaction of
f328865f
MD
323`urcu_<flavor>_synchronize_rcu()` and RCU read-side with mutexes. If
324`urcu_<flavor>_synchronize_rcu()` is called with a mutex held, this
325mutex (or any mutex which has this mutex in its dependency chain) should
326not be acquired from within a RCU read-side critical section.
d589a916
PP
327
328This is especially important to understand in the context of the
329QSBR flavor: a registered reader thread being "online" by
330default should be considered as within a RCU read-side critical
331section unless explicitly put "offline". Therefore, if
f328865f
MD
332`urcu_qsbr_synchronize_rcu()` is called with a mutex held, this mutex,
333as well as any mutex which has this mutex in its dependency chain should
334only be taken when the RCU reader thread is "offline" (this can be
335performed by calling `urcu_qsbr_thread_offline()`).
d589a916
PP
336
337
338### Interaction with `fork()`
339
340Special care must be taken for applications performing `fork()` without
341any following `exec()`. This is caused by the fact that Linux only clones
342the thread calling `fork()`, and thus never replicates any of the other
343parent thread into the child process. Most `liburcu` implementations
344require that all registrations (as reader, `defer_rcu` and `call_rcu`
345threads) should be released before a `fork()` is performed, except for the
346rather common scenario where `fork()` is immediately followed by `exec()` in
347the child process. The only implementation not subject to that rule is
348`liburcu-bp`, which is designed to handle `fork()` by calling
f328865f
MD
349`urcu_bp_before_fork`, `urcu_bp_after_fork_parent` and
350`urcu_bp_after_fork_child`.
351
352Applications that use `urcu_<flavor>_call_rcu()` and that `fork()`
353without doing an immediate `exec()` must take special action. The
354parent must invoke `urcu_<flavor>_call_rcu_before_fork()` before the
355`fork()` and `urcu_<flavor>_call_rcu_after_fork_parent()` after the
356`fork()`. The child process must invoke
357`urcu_<flavor>_call_rcu_after_fork_child()`. Even though these three
358APIs are suitable for passing to `pthread_atfork()`, use of
359`pthread_atfork()` is **STRONGLY DISCOURAGED** for programs calling the
360glibc memory allocator (`malloc()`, `calloc()`, `free()`, ...) within
361`urcu_<flavor>_call_rcu` callbacks. This is due to limitations in the
362way glibc memory allocator handles calls to the memory allocator from
363concurrent threads while the `pthread_atfork()` handlers are executing.
d589a916
PP
364
365Combining e.g.:
366
f328865f
MD
367 - call to `free()` from callbacks executed within
368 `urcu_<flavor>_call_rcu` worker threads,
369 - executing `urcu_<flavor>_call_rcu` atfork handlers within the glibc
370 pthread atfork mechanism,
d589a916
PP
371
372will sometimes trigger interesting process hangs. This usually
373hangs on a memory allocator lock within glibc.
374
375
376### Thread Local Storage (TLS)
377
378Userspace RCU can fall back on `pthread_getspecific()` to emulate
379TLS variables on systems where it is not available. This behavior
380can be forced by specifying `--disable-compiler-tls` as configure
381argument.
382
383
d4e640c0 384### Usage of `DEBUG_RCU` & `--enable-rcu-debug`
d589a916 385
d4e640c0
JR
386By default the library is configured with internal debugging
387self-checks disabled.
388
389For always-on debugging self-checks:
390 ./configure --enable-rcu-debug
391
392For fine grained enabling of debugging self-checks, build
e400ca4a 393userspace-rcu with DEBUG_RCU defined and compile dependent
d4e640c0
JR
394applications with DEBUG_RCU defined when necessary.
395
396Warning: Enabling this feature result in a performance penalty.
d589a916
PP
397
398
399### Usage of `DEBUG_YIELD`
400
401`DEBUG_YIELD` is used to add random delays in the code for testing
402purposes.
403
404
405### SMP support
406
407By default the library is configured to use synchronization primitives
408adequate for SMP systems. On uniprocessor systems, support for SMP
409systems can be disabled with:
410
411 ./configure --disable-smp-support
412
413theoretically yielding slightly better performance.
414
415
d7c76f85
MD
416### Usage of `--enable-cds-lfht-iter-debug`
417
418By default the library is configured with extra debugging checks for
419lock-free hash table iterator traversal disabled.
420
421Building liburcu with --enable-cds-lfht-iter-debug and rebuilding
422application to match the ABI change allows finding cases where the hash
423table iterator is re-purposed to be used on a different hash table while
424still being used to iterate on a hash table.
425
426This option alters the rculfhash ABI. Make sure to compile both library
427and application with matching configuration.
428
429
d589a916
PP
430Make targets
431------------
432
433In addition to the usual `make check` target, Userspace RCU features
434`make regtest` and `make bench` targets:
435
436 - `make check`: short tests, meant to be run when rebuilding or
437 porting Userspace RCU.
438 - `make regtest`: long (many hours) test, meant to be run when
439 modifying Userspace RCU or porting it to a new architecture or
440 operating system.
441 - `make bench`: long (many hours) benchmarks.
442
443
444Contacts
445--------
446
447You can contact the maintainers on the following mailing list:
dcb9c05a 448`lttng-dev@lists.lttng.org`.
This page took 0.085656 seconds and 4 git commands to generate.