ccollins476ad commented on a change in pull request #1223: util/rwlock: Readers–writer lock URL: https://github.com/apache/mynewt-core/pull/1223#discussion_r199240587
########## File path: util/rwlock/src/rwlock.c ########## @@ -0,0 +1,213 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +#include "os/mynewt.h" +#include "rwlock/rwlock.h" + +#if MYNEWT_VAL(RWLOCK_DEBUG) +#define RWLOCK_DBG_ASSERT(expr) (assert(expr)) +#else +#define RWLOCK_DBG_ASSERT(expr) +#endif + +/** + * Unblocks the next pending user. The caller must lock the mutex prior to + * calling this. + */ +static void +rwlock_unblock(struct rwlock *lock) +{ + RWLOCK_DBG_ASSERT(lock->mtx.mu_owner == g_current_task); + RWLOCK_DBG_ASSERT(lock->handoffs == 0); + + /* Give priority to pending writers. */ + if (lock->pending_writers > 0) { + /* Indicate that ownership is being transfered to a single writer. */ + lock->handoffs = 1; + + os_sem_release(&lock->wsem); + lock->pending_writers--; + } else { + /* Indicate that ownership is being transfered to a collection of + * readers. + */ + lock->handoffs = lock->pending_readers; + + while (lock->pending_readers > 0) { + os_sem_release(&lock->rsem); + lock->pending_readers--; + } + } +} + +static void +rwlock_complete_handoff(struct rwlock *lock) +{ + RWLOCK_DBG_ASSERT(lock->mtx.mu_owner == g_current_task); + RWLOCK_DBG_ASSERT(lock->handoffs > 0); + lock->handoffs--; +} + +/** + * Indicates whether a prospective reader must wait for the lock to become + * available. + */ +static bool +rwlock_read_must_block(const struct rwlock *lock) +{ + RWLOCK_DBG_ASSERT(lock->mtx.mu_owner == g_current_task); + + return lock->active_writer || + lock->pending_writers > 0 || + lock->handoffs > 0; +} + +/** + * Indicates whether a prospective writer must wait for the lock to become + * available. + */ +static bool +rwlock_write_must_block(const struct rwlock *lock) +{ + RWLOCK_DBG_ASSERT(lock->mtx.mu_owner == g_current_task); + + return lock->active_writer || + lock->num_readers > 0 || + lock->handoffs > 0; +} + +void +rwlock_acquire_read(struct rwlock *lock) +{ + bool acquired; + + os_mutex_pend(&lock->mtx, OS_TIMEOUT_NEVER); + + if (rwlock_read_must_block(lock)) { + lock->pending_readers++; + acquired = false; + } else { + lock->num_readers++; + acquired = true; + } + + os_mutex_release(&lock->mtx); + + if (acquired) { + /* No contention; lock acquired. */ + return; + } + + /* Wait for the lock to become available. */ + os_sem_pend(&lock->rsem, OS_TIMEOUT_NEVER); + + /* Record reader ownership. */ + os_mutex_pend(&lock->mtx, OS_TIMEOUT_NEVER); + lock->num_readers++; + rwlock_complete_handoff(lock); + os_mutex_release(&lock->mtx); +} + +void +rwlock_release_read(struct rwlock *lock) +{ + os_mutex_pend(&lock->mtx, OS_TIMEOUT_NEVER); + + RWLOCK_DBG_ASSERT(lock->num_readers > 0); + lock->num_readers--; + + /* If this is the last active reader, unblock a pending writer if there is + * one. + */ + if (lock->num_readers == 0) { + rwlock_unblock(lock); + } + + os_mutex_release(&lock->mtx); +} + +void +rwlock_acquire_write(struct rwlock *lock) +{ + bool acquired; + + os_mutex_pend(&lock->mtx, OS_TIMEOUT_NEVER); + + if (rwlock_write_must_block(lock)) { + lock->pending_writers++; + acquired = false; + } else { + lock->active_writer = true; + acquired = true; + } + + os_mutex_release(&lock->mtx); + + if (acquired) { + /* No contention; lock acquired. */ + return; + } + + /* Wait for the lock to become available. */ + os_sem_pend(&lock->wsem, OS_TIMEOUT_NEVER); + + /* Record writer ownership. */ + os_mutex_pend(&lock->mtx, OS_TIMEOUT_NEVER); + lock->active_writer = true; + rwlock_complete_handoff(lock); + os_mutex_release(&lock->mtx); +} + +void +rwlock_release_write(struct rwlock *lock) +{ + os_mutex_pend(&lock->mtx, OS_TIMEOUT_NEVER); + + RWLOCK_DBG_ASSERT(lock->active_writer); + lock->active_writer = false; + + rwlock_unblock(lock); + + os_mutex_release(&lock->mtx); +} + +int +rwlock_init(struct rwlock *lock) +{ + int rc; + + *lock = (struct rwlock) { 0 }; Review comment: Wow, that would be quite a regression for Apple to introduce. I have an equally recent version of clang on my linux machine (5.0.2) which doesn't have this bug. I wonder if it is correct to change code to accommodate a broken version of a compiler... ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
