171 lines · plain
1=======================2Generic Mutex Subsystem3=======================4 5started by Ingo Molnar <mingo@redhat.com>6 7updated by Davidlohr Bueso <davidlohr@hp.com>8 9What are mutexes?10-----------------11 12In the Linux kernel, mutexes refer to a particular locking primitive13that enforces serialization on shared memory systems, and not only to14the generic term referring to 'mutual exclusion' found in academia15or similar theoretical text books. Mutexes are sleeping locks which16behave similarly to binary semaphores, and were introduced in 2006[1]17as an alternative to these. This new data structure provided a number18of advantages, including simpler interfaces, and at that time smaller19code (see Disadvantages).20 21[1] https://lwn.net/Articles/164802/22 23Implementation24--------------25 26Mutexes are represented by 'struct mutex', defined in include/linux/mutex.h27and implemented in kernel/locking/mutex.c. These locks use an atomic variable28(->owner) to keep track of the lock state during its lifetime. Field owner29actually contains `struct task_struct *` to the current lock owner and it is30therefore NULL if not currently owned. Since task_struct pointers are aligned31to at least L1_CACHE_BYTES, low bits (3) are used to store extra state (e.g.,32if waiter list is non-empty). In its most basic form it also includes a33wait-queue and a spinlock that serializes access to it. Furthermore,34CONFIG_MUTEX_SPIN_ON_OWNER=y systems use a spinner MCS lock (->osq), described35below in (ii).36 37When acquiring a mutex, there are three possible paths that can be38taken, depending on the state of the lock:39 40(i) fastpath: tries to atomically acquire the lock by cmpxchg()ing the owner with41 the current task. This only works in the uncontended case (cmpxchg() checks42 against 0UL, so all 3 state bits above have to be 0). If the lock is43 contended it goes to the next possible path.44 45(ii) midpath: aka optimistic spinning, tries to spin for acquisition46 while the lock owner is running and there are no other tasks ready47 to run that have higher priority (need_resched). The rationale is48 that if the lock owner is running, it is likely to release the lock49 soon. The mutex spinners are queued up using MCS lock so that only50 one spinner can compete for the mutex.51 52 The MCS lock (proposed by Mellor-Crummey and Scott) is a simple spinlock53 with the desirable properties of being fair and with each cpu trying54 to acquire the lock spinning on a local variable. It avoids expensive55 cacheline bouncing that common test-and-set spinlock implementations56 incur. An MCS-like lock is specially tailored for optimistic spinning57 for sleeping lock implementation. An important feature of the customized58 MCS lock is that it has the extra property that spinners are able to exit59 the MCS spinlock queue when they need to reschedule. This further helps60 avoid situations where MCS spinners that need to reschedule would continue61 waiting to spin on mutex owner, only to go directly to slowpath upon62 obtaining the MCS lock.63 64 65(iii) slowpath: last resort, if the lock is still unable to be acquired,66 the task is added to the wait-queue and sleeps until woken up by the67 unlock path. Under normal circumstances it blocks as TASK_UNINTERRUPTIBLE.68 69While formally kernel mutexes are sleepable locks, it is path (ii) that70makes them more practically a hybrid type. By simply not interrupting a71task and busy-waiting for a few cycles instead of immediately sleeping,72the performance of this lock has been seen to significantly improve a73number of workloads. Note that this technique is also used for rw-semaphores.74 75Semantics76---------77 78The mutex subsystem checks and enforces the following rules:79 80 - Only one task can hold the mutex at a time.81 - Only the owner can unlock the mutex.82 - Multiple unlocks are not permitted.83 - Recursive locking/unlocking is not permitted.84 - A mutex must only be initialized via the API (see below).85 - A task may not exit with a mutex held.86 - Memory areas where held locks reside must not be freed.87 - Held mutexes must not be reinitialized.88 - Mutexes may not be used in hardware or software interrupt89 contexts such as tasklets and timers.90 91These semantics are fully enforced when CONFIG DEBUG_MUTEXES is enabled.92In addition, the mutex debugging code also implements a number of other93features that make lock debugging easier and faster:94 95 - Uses symbolic names of mutexes, whenever they are printed96 in debug output.97 - Point-of-acquire tracking, symbolic lookup of function names,98 list of all locks held in the system, printout of them.99 - Owner tracking.100 - Detects self-recursing locks and prints out all relevant info.101 - Detects multi-task circular deadlocks and prints out all affected102 locks and tasks (and only those tasks).103 104Mutexes - and most other sleeping locks like rwsems - do not provide an105implicit reference for the memory they occupy, which reference is released106with mutex_unlock().107 108[ This is in contrast with spin_unlock() [or completion_done()], which109 APIs can be used to guarantee that the memory is not touched by the110 lock implementation after spin_unlock()/completion_done() releases111 the lock. ]112 113mutex_unlock() may access the mutex structure even after it has internally114released the lock already - so it's not safe for another context to115acquire the mutex and assume that the mutex_unlock() context is not using116the structure anymore.117 118The mutex user must ensure that the mutex is not destroyed while a119release operation is still in progress - in other words, callers of120mutex_unlock() must ensure that the mutex stays alive until mutex_unlock()121has returned.122 123Interfaces124----------125Statically define the mutex::126 127 DEFINE_MUTEX(name);128 129Dynamically initialize the mutex::130 131 mutex_init(mutex);132 133Acquire the mutex, uninterruptible::134 135 void mutex_lock(struct mutex *lock);136 void mutex_lock_nested(struct mutex *lock, unsigned int subclass);137 int mutex_trylock(struct mutex *lock);138 139Acquire the mutex, interruptible::140 141 int mutex_lock_interruptible_nested(struct mutex *lock,142 unsigned int subclass);143 int mutex_lock_interruptible(struct mutex *lock);144 145Acquire the mutex, interruptible, if dec to 0::146 147 int atomic_dec_and_mutex_lock(atomic_t *cnt, struct mutex *lock);148 149Unlock the mutex::150 151 void mutex_unlock(struct mutex *lock);152 153Test if the mutex is taken::154 155 int mutex_is_locked(struct mutex *lock);156 157Disadvantages158-------------159 160Unlike its original design and purpose, 'struct mutex' is among the largest161locks in the kernel. E.g: on x86-64 it is 32 bytes, where 'struct semaphore'162is 24 bytes and rw_semaphore is 40 bytes. Larger structure sizes mean more CPU163cache and memory footprint.164 165When to use mutexes166-------------------167 168Unless the strict semantics of mutexes are unsuitable and/or the critical169region prevents the lock from being shared, always prefer them to any other170locking primitive.171