| 1 | /**************************************************************************** |
| 2 | ** |
| 3 | ** Copyright (C) 2017 The Qt Company Ltd. |
| 4 | ** Copyright (C) 2018 Intel Corporation. |
| 5 | ** Contact: https://www.qt.io/licensing/ |
| 6 | ** |
| 7 | ** This file is part of the QtCore module of the Qt Toolkit. |
| 8 | ** |
| 9 | ** $QT_BEGIN_LICENSE:LGPL$ |
| 10 | ** Commercial License Usage |
| 11 | ** Licensees holding valid commercial Qt licenses may use this file in |
| 12 | ** accordance with the commercial license agreement provided with the |
| 13 | ** Software or, alternatively, in accordance with the terms contained in |
| 14 | ** a written agreement between you and The Qt Company. For licensing terms |
| 15 | ** and conditions see https://www.qt.io/terms-conditions. For further |
| 16 | ** information use the contact form at https://www.qt.io/contact-us. |
| 17 | ** |
| 18 | ** GNU Lesser General Public License Usage |
| 19 | ** Alternatively, this file may be used under the terms of the GNU Lesser |
| 20 | ** General Public License version 3 as published by the Free Software |
| 21 | ** Foundation and appearing in the file LICENSE.LGPL3 included in the |
| 22 | ** packaging of this file. Please review the following information to |
| 23 | ** ensure the GNU Lesser General Public License version 3 requirements |
| 24 | ** will be met: https://www.gnu.org/licenses/lgpl-3.0.html. |
| 25 | ** |
| 26 | ** GNU General Public License Usage |
| 27 | ** Alternatively, this file may be used under the terms of the GNU |
| 28 | ** General Public License version 2.0 or (at your option) the GNU General |
| 29 | ** Public license version 3 or any later version approved by the KDE Free |
| 30 | ** Qt Foundation. The licenses are as published by the Free Software |
| 31 | ** Foundation and appearing in the file LICENSE.GPL2 and LICENSE.GPL3 |
| 32 | ** included in the packaging of this file. Please review the following |
| 33 | ** information to ensure the GNU General Public License requirements will |
| 34 | ** be met: https://www.gnu.org/licenses/gpl-2.0.html and |
| 35 | ** https://www.gnu.org/licenses/gpl-3.0.html. |
| 36 | ** |
| 37 | ** $QT_END_LICENSE$ |
| 38 | ** |
| 39 | ****************************************************************************/ |
| 40 | |
| 41 | #include "qsemaphore.h" |
| 42 | #include "qmutex.h" |
| 43 | #include "qfutex_p.h" |
| 44 | #include "qwaitcondition.h" |
| 45 | #include "qdeadlinetimer.h" |
| 46 | #include "qdatetime.h" |
| 47 | |
| 48 | QT_BEGIN_NAMESPACE |
| 49 | |
| 50 | using namespace QtFutex; |
| 51 | |
| 52 | /*! |
| 53 | \class QSemaphore |
| 54 | \inmodule QtCore |
| 55 | \brief The QSemaphore class provides a general counting semaphore. |
| 56 | |
| 57 | \threadsafe |
| 58 | |
| 59 | \ingroup thread |
| 60 | |
| 61 | A semaphore is a generalization of a mutex. While a mutex can |
| 62 | only be locked once, it's possible to acquire a semaphore |
| 63 | multiple times. Semaphores are typically used to protect a |
| 64 | certain number of identical resources. |
| 65 | |
| 66 | Semaphores support two fundamental operations, acquire() and |
| 67 | release(): |
| 68 | |
| 69 | \list |
| 70 | \li acquire(\e{n}) tries to acquire \e n resources. If there aren't |
| 71 | that many resources available, the call will block until this |
| 72 | is the case. |
| 73 | \li release(\e{n}) releases \e n resources. |
| 74 | \endlist |
| 75 | |
| 76 | There's also a tryAcquire() function that returns immediately if |
| 77 | it cannot acquire the resources, and an available() function that |
| 78 | returns the number of available resources at any time. |
| 79 | |
| 80 | Example: |
| 81 | |
| 82 | \snippet code/src_corelib_thread_qsemaphore.cpp 0 |
| 83 | |
| 84 | A typical application of semaphores is for controlling access to |
| 85 | a circular buffer shared by a producer thread and a consumer |
| 86 | thread. The \l{Semaphores Example} shows how |
| 87 | to use QSemaphore to solve that problem. |
| 88 | |
| 89 | A non-computing example of a semaphore would be dining at a |
| 90 | restaurant. A semaphore is initialized with the number of chairs |
| 91 | in the restaurant. As people arrive, they want a seat. As seats |
| 92 | are filled, available() is decremented. As people leave, the |
| 93 | available() is incremented, allowing more people to enter. If a |
| 94 | party of 10 people want to be seated, but there are only 9 seats, |
| 95 | those 10 people will wait, but a party of 4 people would be |
| 96 | seated (taking the available seats to 5, making the party of 10 |
| 97 | people wait longer). |
| 98 | |
| 99 | \sa QSemaphoreReleaser, QMutex, QWaitCondition, QThread, {Semaphores Example} |
| 100 | */ |
| 101 | |
| 102 | /* |
| 103 | QSemaphore futex operation |
| 104 | |
| 105 | QSemaphore stores a 32-bit integer with the counter of currently available |
| 106 | tokens (value between 0 and INT_MAX). When a thread attempts to acquire n |
| 107 | tokens and the counter is larger than that, we perform a compare-and-swap |
| 108 | with the new count. If that succeeds, the acquisition worked; if not, we |
| 109 | loop again because the counter changed. If there were not enough tokens, |
| 110 | we'll perform a futex-wait. |
| 111 | |
| 112 | Before we do, we set the high bit in the futex to indicate that semaphore |
| 113 | is contended: that is, there's a thread waiting for more tokens. On |
| 114 | release() for n tokens, we perform a fetch-and-add of n and then check if |
| 115 | that high bit was set. If it was, then we clear that bit and perform a |
| 116 | futex-wake on the semaphore to indicate the waiting threads can wake up and |
| 117 | acquire tokens. Which ones get woken up is unspecified. |
| 118 | |
| 119 | If the system has the ability to wake up a precise number of threads, has |
| 120 | Linux's FUTEX_WAKE_OP functionality, and is 64-bit, instead of using a |
| 121 | single bit indicating a contended semaphore, we'll store the number of |
| 122 | tokens *plus* total number of waiters in the high word. Additionally, all |
| 123 | multi-token waiters will be waiting on that high word. So when releasing n |
| 124 | tokens on those systems, we tell the kernel to wake up n single-token |
| 125 | threads and all of the multi-token ones. Which threads get woken up is |
| 126 | unspecified, but it's likely single-token threads will get woken up first. |
| 127 | */ |
| 128 | |
| 129 | #if defined(FUTEX_OP) && QT_POINTER_SIZE > 4 |
| 130 | static Q_CONSTEXPR bool futexHasWaiterCount = true; |
| 131 | #else |
| 132 | static Q_CONSTEXPR bool futexHasWaiterCount = false; |
| 133 | #endif |
| 134 | |
| 135 | static const quintptr futexNeedsWakeAllBit = |
| 136 | Q_UINT64_C(1) << (sizeof(quintptr) * CHAR_BIT - 1); |
| 137 | |
| 138 | static int futexAvailCounter(quintptr v) |
| 139 | { |
| 140 | // the low 31 bits |
| 141 | if (futexHasWaiterCount) { |
| 142 | // the high bit of the low word isn't used |
| 143 | Q_ASSERT((v & 0x80000000U) == 0); |
| 144 | |
| 145 | // so we can be a little faster |
| 146 | return int(unsigned(v)); |
| 147 | } |
| 148 | return int(v & 0x7fffffffU); |
| 149 | } |
| 150 | |
| 151 | static bool futexNeedsWake(quintptr v) |
| 152 | { |
| 153 | // If we're counting waiters, the number of waiters is stored in the low 31 |
| 154 | // bits of the high word (that is, bits 32-62). If we're not, then we use |
| 155 | // bit 31 to indicate anyone is waiting. Either way, if any bit 31 or above |
| 156 | // is set, there are waiters. |
| 157 | return v >> 31; |
| 158 | } |
| 159 | |
| 160 | static QBasicAtomicInteger<quint32> *futexLow32(QBasicAtomicInteger<quintptr> *ptr) |
| 161 | { |
| 162 | auto result = reinterpret_cast<QBasicAtomicInteger<quint32> *>(ptr); |
| 163 | #if Q_BYTE_ORDER == Q_BIG_ENDIAN && QT_POINTER_SIZE > 4 |
| 164 | ++result; |
| 165 | #endif |
| 166 | return result; |
| 167 | } |
| 168 | |
| 169 | static QBasicAtomicInteger<quint32> *futexHigh32(QBasicAtomicInteger<quintptr> *ptr) |
| 170 | { |
| 171 | auto result = reinterpret_cast<QBasicAtomicInteger<quint32> *>(ptr); |
| 172 | #if Q_BYTE_ORDER == Q_LITTLE_ENDIAN && QT_POINTER_SIZE > 4 |
| 173 | ++result; |
| 174 | #endif |
| 175 | return result; |
| 176 | } |
| 177 | |
| 178 | template <bool IsTimed> bool |
| 179 | futexSemaphoreTryAcquire_loop(QBasicAtomicInteger<quintptr> &u, quintptr curValue, quintptr nn, int timeout) |
| 180 | { |
| 181 | QDeadlineTimer timer(IsTimed ? QDeadlineTimer(timeout) : QDeadlineTimer()); |
| 182 | qint64 remainingTime = timeout * Q_INT64_C(1000) * 1000; |
| 183 | int n = int(unsigned(nn)); |
| 184 | |
| 185 | // we're called after one testAndSet, so start by waiting first |
| 186 | goto start_wait; |
| 187 | |
| 188 | forever { |
| 189 | if (futexAvailCounter(v: curValue) >= n) { |
| 190 | // try to acquire |
| 191 | quintptr newValue = curValue - nn; |
| 192 | if (u.testAndSetOrdered(expectedValue: curValue, newValue, currentValue&: curValue)) |
| 193 | return true; // succeeded! |
| 194 | continue; |
| 195 | } |
| 196 | |
| 197 | // not enough tokens available, put us to wait |
| 198 | if (remainingTime == 0) |
| 199 | return false; |
| 200 | |
| 201 | // indicate we're waiting |
| 202 | start_wait: |
| 203 | auto ptr = futexLow32(ptr: &u); |
| 204 | if (n > 1 || !futexHasWaiterCount) { |
| 205 | u.fetchAndOrRelaxed(valueToAdd: futexNeedsWakeAllBit); |
| 206 | curValue |= futexNeedsWakeAllBit; |
| 207 | if (n > 1 && futexHasWaiterCount) { |
| 208 | ptr = futexHigh32(ptr: &u); |
| 209 | //curValue >>= 32; // but this is UB in 32-bit, so roundabout: |
| 210 | curValue = quint64(curValue) >> 32; |
| 211 | } |
| 212 | } |
| 213 | |
| 214 | if (IsTimed && remainingTime > 0) { |
| 215 | bool timedout = !futexWait(futex&: *ptr, expectedValue: curValue, nstimeout: remainingTime); |
| 216 | if (timedout) |
| 217 | return false; |
| 218 | } else { |
| 219 | futexWait(futex&: *ptr, expectedValue: curValue); |
| 220 | } |
| 221 | |
| 222 | curValue = u.loadAcquire(); |
| 223 | if (IsTimed) |
| 224 | remainingTime = timer.remainingTimeNSecs(); |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | template <bool IsTimed> bool futexSemaphoreTryAcquire(QBasicAtomicInteger<quintptr> &u, int n, int timeout) |
| 229 | { |
| 230 | // Try to acquire without waiting (we still loop because the testAndSet |
| 231 | // call can fail). |
| 232 | quintptr nn = unsigned(n); |
| 233 | if (futexHasWaiterCount) |
| 234 | nn |= quint64(nn) << 32; // token count replicated in high word |
| 235 | |
| 236 | quintptr curValue = u.loadAcquire(); |
| 237 | while (futexAvailCounter(v: curValue) >= n) { |
| 238 | // try to acquire |
| 239 | quintptr newValue = curValue - nn; |
| 240 | if (u.testAndSetOrdered(expectedValue: curValue, newValue, currentValue&: curValue)) |
| 241 | return true; // succeeded! |
| 242 | } |
| 243 | if (timeout == 0) |
| 244 | return false; |
| 245 | |
| 246 | // we need to wait |
| 247 | quintptr oneWaiter = quintptr(Q_UINT64_C(1) << 32); // zero on 32-bit |
| 248 | if (futexHasWaiterCount) { |
| 249 | // increase the waiter count |
| 250 | u.fetchAndAddRelaxed(valueToAdd: oneWaiter); |
| 251 | |
| 252 | // We don't use the fetched value from above so futexWait() fails if |
| 253 | // it changed after the testAndSetOrdered above. |
| 254 | if ((quint64(curValue) >> 32) == 0x7fffffff) |
| 255 | return false; // overflow! |
| 256 | curValue += oneWaiter; |
| 257 | |
| 258 | // Also adjust nn to subtract oneWaiter when we succeed in acquiring. |
| 259 | nn += oneWaiter; |
| 260 | } |
| 261 | |
| 262 | if (futexSemaphoreTryAcquire_loop<IsTimed>(u, curValue, nn, timeout)) |
| 263 | return true; |
| 264 | |
| 265 | if (futexHasWaiterCount) { |
| 266 | // decrement the number of threads waiting |
| 267 | Q_ASSERT(futexHigh32(&u)->loadRelaxed() & 0x7fffffffU); |
| 268 | u.fetchAndSubRelaxed(valueToAdd: oneWaiter); |
| 269 | } |
| 270 | return false; |
| 271 | } |
| 272 | |
| 273 | class QSemaphorePrivate { |
| 274 | public: |
| 275 | inline QSemaphorePrivate(int n) : avail(n) { } |
| 276 | |
| 277 | QMutex mutex; |
| 278 | QWaitCondition cond; |
| 279 | |
| 280 | int avail; |
| 281 | }; |
| 282 | |
| 283 | /*! |
| 284 | Creates a new semaphore and initializes the number of resources |
| 285 | it guards to \a n (by default, 0). |
| 286 | |
| 287 | \sa release(), available() |
| 288 | */ |
| 289 | QSemaphore::QSemaphore(int n) |
| 290 | { |
| 291 | Q_ASSERT_X(n >= 0, "QSemaphore" , "parameter 'n' must be non-negative" ); |
| 292 | if (futexAvailable()) { |
| 293 | quintptr nn = unsigned(n); |
| 294 | if (futexHasWaiterCount) |
| 295 | nn |= quint64(nn) << 32; // token count replicated in high word |
| 296 | u.storeRelaxed(newValue: nn); |
| 297 | } else { |
| 298 | d = new QSemaphorePrivate(n); |
| 299 | } |
| 300 | } |
| 301 | |
| 302 | /*! |
| 303 | Destroys the semaphore. |
| 304 | |
| 305 | \warning Destroying a semaphore that is in use may result in |
| 306 | undefined behavior. |
| 307 | */ |
| 308 | QSemaphore::~QSemaphore() |
| 309 | { |
| 310 | if (!futexAvailable()) |
| 311 | delete d; |
| 312 | } |
| 313 | |
| 314 | /*! |
| 315 | Tries to acquire \c n resources guarded by the semaphore. If \a n |
| 316 | > available(), this call will block until enough resources are |
| 317 | available. |
| 318 | |
| 319 | \sa release(), available(), tryAcquire() |
| 320 | */ |
| 321 | void QSemaphore::acquire(int n) |
| 322 | { |
| 323 | Q_ASSERT_X(n >= 0, "QSemaphore::acquire" , "parameter 'n' must be non-negative" ); |
| 324 | |
| 325 | if (futexAvailable()) { |
| 326 | futexSemaphoreTryAcquire<false>(u&: u, n, timeout: -1); |
| 327 | return; |
| 328 | } |
| 329 | |
| 330 | QMutexLocker locker(&d->mutex); |
| 331 | while (n > d->avail) |
| 332 | d->cond.wait(lockedMutex: locker.mutex()); |
| 333 | d->avail -= n; |
| 334 | } |
| 335 | |
| 336 | /*! |
| 337 | Releases \a n resources guarded by the semaphore. |
| 338 | |
| 339 | This function can be used to "create" resources as well. For |
| 340 | example: |
| 341 | |
| 342 | \snippet code/src_corelib_thread_qsemaphore.cpp 1 |
| 343 | |
| 344 | QSemaphoreReleaser is a \l{http://en.cppreference.com/w/cpp/language/raii}{RAII} |
| 345 | wrapper around this function. |
| 346 | |
| 347 | \sa acquire(), available(), QSemaphoreReleaser |
| 348 | */ |
| 349 | void QSemaphore::release(int n) |
| 350 | { |
| 351 | Q_ASSERT_X(n >= 0, "QSemaphore::release" , "parameter 'n' must be non-negative" ); |
| 352 | |
| 353 | if (futexAvailable()) { |
| 354 | quintptr nn = unsigned(n); |
| 355 | if (futexHasWaiterCount) |
| 356 | nn |= quint64(nn) << 32; // token count replicated in high word |
| 357 | quintptr prevValue = u.fetchAndAddRelease(valueToAdd: nn); |
| 358 | if (futexNeedsWake(v: prevValue)) { |
| 359 | #ifdef FUTEX_OP |
| 360 | if (futexHasWaiterCount) { |
| 361 | /* |
| 362 | On 64-bit systems, the single-token waiters wait on the low half |
| 363 | and the multi-token waiters wait on the upper half. So we ask |
| 364 | the kernel to wake up n single-token waiters and all multi-token |
| 365 | waiters (if any), and clear the multi-token wait bit. |
| 366 | |
| 367 | atomic { |
| 368 | int oldval = *upper; |
| 369 | *upper = oldval | 0; |
| 370 | futexWake(lower, n); |
| 371 | if (oldval != 0) // always true |
| 372 | futexWake(upper, INT_MAX); |
| 373 | } |
| 374 | */ |
| 375 | quint32 op = FUTEX_OP_OR; |
| 376 | quint32 oparg = 0; |
| 377 | quint32 cmp = FUTEX_OP_CMP_NE; |
| 378 | quint32 cmparg = 0; |
| 379 | u.fetchAndAndRelease(valueToAdd: futexNeedsWakeAllBit - 1); |
| 380 | futexWakeOp(futex1&: *futexLow32(ptr: &u), wake1: n, INT_MAX, futex2&: *futexHigh32(ptr: &u), FUTEX_OP(op, oparg, cmp, cmparg)); |
| 381 | return; |
| 382 | } |
| 383 | #endif |
| 384 | // Unset the bit and wake everyone. There are two possibilities |
| 385 | // under which a thread can set the bit between the AND and the |
| 386 | // futexWake: |
| 387 | // 1) it did see the new counter value, but it wasn't enough for |
| 388 | // its acquisition anyway, so it has to wait; |
| 389 | // 2) it did not see the new counter value, in which case its |
| 390 | // futexWait will fail. |
| 391 | u.fetchAndAndRelease(valueToAdd: futexNeedsWakeAllBit - 1); |
| 392 | if (futexHasWaiterCount) { |
| 393 | futexWakeAll(futex&: *futexLow32(ptr: &u)); |
| 394 | futexWakeAll(futex&: *futexHigh32(ptr: &u)); |
| 395 | } else { |
| 396 | futexWakeAll(futex&: u); |
| 397 | } |
| 398 | } |
| 399 | return; |
| 400 | } |
| 401 | |
| 402 | QMutexLocker locker(&d->mutex); |
| 403 | d->avail += n; |
| 404 | d->cond.wakeAll(); |
| 405 | } |
| 406 | |
| 407 | /*! |
| 408 | Returns the number of resources currently available to the |
| 409 | semaphore. This number can never be negative. |
| 410 | |
| 411 | \sa acquire(), release() |
| 412 | */ |
| 413 | int QSemaphore::available() const |
| 414 | { |
| 415 | if (futexAvailable()) |
| 416 | return futexAvailCounter(v: u.loadRelaxed()); |
| 417 | |
| 418 | QMutexLocker locker(&d->mutex); |
| 419 | return d->avail; |
| 420 | } |
| 421 | |
| 422 | /*! |
| 423 | Tries to acquire \c n resources guarded by the semaphore and |
| 424 | returns \c true on success. If available() < \a n, this call |
| 425 | immediately returns \c false without acquiring any resources. |
| 426 | |
| 427 | Example: |
| 428 | |
| 429 | \snippet code/src_corelib_thread_qsemaphore.cpp 2 |
| 430 | |
| 431 | \sa acquire() |
| 432 | */ |
| 433 | bool QSemaphore::tryAcquire(int n) |
| 434 | { |
| 435 | Q_ASSERT_X(n >= 0, "QSemaphore::tryAcquire" , "parameter 'n' must be non-negative" ); |
| 436 | |
| 437 | if (futexAvailable()) |
| 438 | return futexSemaphoreTryAcquire<false>(u&: u, n, timeout: 0); |
| 439 | |
| 440 | QMutexLocker locker(&d->mutex); |
| 441 | if (n > d->avail) |
| 442 | return false; |
| 443 | d->avail -= n; |
| 444 | return true; |
| 445 | } |
| 446 | |
| 447 | /*! |
| 448 | Tries to acquire \c n resources guarded by the semaphore and |
| 449 | returns \c true on success. If available() < \a n, this call will |
| 450 | wait for at most \a timeout milliseconds for resources to become |
| 451 | available. |
| 452 | |
| 453 | Note: Passing a negative number as the \a timeout is equivalent to |
| 454 | calling acquire(), i.e. this function will wait forever for |
| 455 | resources to become available if \a timeout is negative. |
| 456 | |
| 457 | Example: |
| 458 | |
| 459 | \snippet code/src_corelib_thread_qsemaphore.cpp 3 |
| 460 | |
| 461 | \sa acquire() |
| 462 | */ |
| 463 | bool QSemaphore::tryAcquire(int n, int timeout) |
| 464 | { |
| 465 | Q_ASSERT_X(n >= 0, "QSemaphore::tryAcquire" , "parameter 'n' must be non-negative" ); |
| 466 | |
| 467 | // We're documented to accept any negative value as "forever" |
| 468 | // but QDeadlineTimer only accepts -1. |
| 469 | timeout = qMax(a: timeout, b: -1); |
| 470 | |
| 471 | if (futexAvailable()) |
| 472 | return futexSemaphoreTryAcquire<true>(u&: u, n, timeout); |
| 473 | |
| 474 | QDeadlineTimer timer(timeout); |
| 475 | QMutexLocker locker(&d->mutex); |
| 476 | while (n > d->avail && !timer.hasExpired()) { |
| 477 | if (!d->cond.wait(lockedMutex: locker.mutex(), deadline: timer)) |
| 478 | return false; |
| 479 | } |
| 480 | if (n > d->avail) |
| 481 | return false; |
| 482 | d->avail -= n; |
| 483 | return true; |
| 484 | |
| 485 | |
| 486 | } |
| 487 | |
| 488 | /*! |
| 489 | \class QSemaphoreReleaser |
| 490 | \brief The QSemaphoreReleaser class provides exception-safe deferral of a QSemaphore::release() call. |
| 491 | \since 5.10 |
| 492 | \ingroup thread |
| 493 | \inmodule QtCore |
| 494 | |
| 495 | \reentrant |
| 496 | |
| 497 | QSemaphoreReleaser can be used wherever you would otherwise use |
| 498 | QSemaphore::release(). Constructing a QSemaphoreReleaser defers the |
| 499 | release() call on the semaphore until the QSemaphoreReleaser is |
| 500 | destroyed (see |
| 501 | \l{http://en.cppreference.com/w/cpp/language/raii}{RAII pattern}). |
| 502 | |
| 503 | You can use this to reliably release a semaphore to avoid dead-lock |
| 504 | in the face of exceptions or early returns: |
| 505 | |
| 506 | \snippet code/src_corelib_thread_qsemaphore.cpp 4 |
| 507 | |
| 508 | If an early return is taken or an exception is thrown before the |
| 509 | \c{sem.release()} call is reached, the semaphore is not released, |
| 510 | possibly preventing the thread waiting in the corresponding |
| 511 | \c{sem.acquire()} call from ever continuing execution. |
| 512 | |
| 513 | When using RAII instead: |
| 514 | |
| 515 | \snippet code/src_corelib_thread_qsemaphore.cpp 5 |
| 516 | |
| 517 | this can no longer happen, because the compiler will make sure that |
| 518 | the QSemaphoreReleaser destructor is always called, and therefore |
| 519 | the semaphore is always released. |
| 520 | |
| 521 | QSemaphoreReleaser is move-enabled and can therefore be returned |
| 522 | from functions to transfer responsibility for releasing a semaphore |
| 523 | out of a function or a scope: |
| 524 | |
| 525 | \snippet code/src_corelib_thread_qsemaphore.cpp 6 |
| 526 | |
| 527 | A QSemaphoreReleaser can be canceled by a call to cancel(). A canceled |
| 528 | semaphore releaser will no longer call QSemaphore::release() in its |
| 529 | destructor. |
| 530 | |
| 531 | \sa QMutexLocker |
| 532 | */ |
| 533 | |
| 534 | /*! |
| 535 | \fn QSemaphoreReleaser::QSemaphoreReleaser() |
| 536 | |
| 537 | Default constructor. Creates a QSemaphoreReleaser that does nothing. |
| 538 | */ |
| 539 | |
| 540 | /*! |
| 541 | \fn QSemaphoreReleaser::QSemaphoreReleaser(QSemaphore &sem, int n) |
| 542 | |
| 543 | Constructor. Stores the arguments and calls \a{sem}.release(\a{n}) |
| 544 | in the destructor. |
| 545 | */ |
| 546 | |
| 547 | /*! |
| 548 | \fn QSemaphoreReleaser::QSemaphoreReleaser(QSemaphore *sem, int n) |
| 549 | |
| 550 | Constructor. Stores the arguments and calls \a{sem}->release(\a{n}) |
| 551 | in the destructor. |
| 552 | */ |
| 553 | |
| 554 | /*! |
| 555 | \fn QSemaphoreReleaser::QSemaphoreReleaser(QSemaphoreReleaser &&other) |
| 556 | |
| 557 | Move constructor. Takes over responsibility to call QSemaphore::release() |
| 558 | from \a other, which in turn is canceled. |
| 559 | |
| 560 | \sa cancel() |
| 561 | */ |
| 562 | |
| 563 | /*! |
| 564 | \fn QSemaphoreReleaser::operator=(QSemaphoreReleaser &&other) |
| 565 | |
| 566 | Move assignment operator. Takes over responsibility to call QSemaphore::release() |
| 567 | from \a other, which in turn is canceled. |
| 568 | |
| 569 | If this semaphore releaser had the responsibility to call some QSemaphore::release() |
| 570 | itself, it performs the call before taking over from \a other. |
| 571 | |
| 572 | \sa cancel() |
| 573 | */ |
| 574 | |
| 575 | /*! |
| 576 | \fn QSemaphoreReleaser::~QSemaphoreReleaser() |
| 577 | |
| 578 | Unless canceled, calls QSemaphore::release() with the arguments provided |
| 579 | to the constructor, or by the last move assignment. |
| 580 | */ |
| 581 | |
| 582 | /*! |
| 583 | \fn QSemaphoreReleaser::swap(QSemaphoreReleaser &other) |
| 584 | |
| 585 | Exchanges the responsibilites of \c{*this} and \a other. |
| 586 | |
| 587 | Unlike move assignment, neither of the two objects ever releases its |
| 588 | semaphore, if any, as a consequence of swapping. |
| 589 | |
| 590 | Therefore this function is very fast and never fails. |
| 591 | */ |
| 592 | |
| 593 | /*! |
| 594 | \fn QSemaphoreReleaser::semaphore() const |
| 595 | |
| 596 | Returns a pointer to the QSemaphore object provided to the constructor, |
| 597 | or by the last move assignment, if any. Otherwise, returns \nullptr. |
| 598 | */ |
| 599 | |
| 600 | /*! |
| 601 | \fn QSemaphoreReleaser::cancel() |
| 602 | |
| 603 | Cancels this QSemaphoreReleaser such that the destructor will no longer |
| 604 | call \c{semaphore()->release()}. Returns the value of semaphore() |
| 605 | before this call. After this call, semaphore() will return \nullptr. |
| 606 | |
| 607 | To enable again, assign a new QSemaphoreReleaser: |
| 608 | |
| 609 | \snippet code/src_corelib_thread_qsemaphore.cpp 7 |
| 610 | */ |
| 611 | |
| 612 | |
| 613 | QT_END_NAMESPACE |
| 614 | |