Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753083AbaBJUld (ORCPT ); Mon, 10 Feb 2014 15:41:33 -0500 Received: from merlin.infradead.org ([205.233.59.134]:35702 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752769AbaBJUl2 (ORCPT ); Mon, 10 Feb 2014 15:41:28 -0500 Message-Id: <20140210203659.671232531@infradead.org> User-Agent: quilt/0.60-1 Date: Mon, 10 Feb 2014 20:58:25 +0100 From: Peter Zijlstra To: linux-kernel@vger.kernel.org Cc: Jason Low , Waiman Long , Peter Zijlstra , mingo@kernel.org, paulmck@linux.vnet.ibm.com, torvalds@linux-foundation.org, tglx@linutronix.de, riel@redhat.com, akpm@linux-foundation.org, davidlohr@hp.com, hpa@zytor.com, andi@firstfloor.org, aswin@hp.com, scott.norton@hp.com, chegu_vinod@hp.com Subject: [PATCH 5/8] locking, mutex: Cancelable MCS lock for adaptive spinning References: <20140210195820.834693028@infradead.org> Content-Disposition: inline; filename=peter_zijlstra-_patch_v2_5_5__mutex-give_spinners_a_chance_to.patch Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Since we want a task waiting for a mutex_lock() to go to sleep and reschedule on need_resched() we must be able to abort the mcs_spin_lock() around the adaptive spin. Therefore implement a cancelable mcs lock. Cc: riel@redhat.com Cc: akpm@linux-foundation.org Cc: davidlohr@hp.com Cc: hpa@zytor.com Cc: andi@firstfloor.org Cc: aswin@hp.com Cc: scott.norton@hp.com Cc: Jason Low Cc: chegu_vinod@hp.com Cc: mingo@kernel.org Cc: paulmck@linux.vnet.ibm.com Cc: Waiman.Long@hp.com Cc: torvalds@linux-foundation.org Cc: tglx@linutronix.de Signed-off-by: Peter Zijlstra --- include/linux/mutex.h | 4 - kernel/locking/Makefile | 2 kernel/locking/mcs_spinlock.c | 167 ++++++++++++++++++++++++++++++++++++++++++ kernel/locking/mcs_spinlock.h | 15 +++ kernel/locking/mutex.c | 10 +- 5 files changed, 191 insertions(+), 7 deletions(-) --- a/include/linux/mutex.h +++ b/include/linux/mutex.h @@ -46,7 +46,7 @@ * - detects multi-task circular deadlocks and prints out all affected * locks and tasks (and only those tasks) */ -struct mcs_spinlock; +struct optimistic_spin_queue; struct mutex { /* 1: unlocked, 0: locked, negative: locked, possible waiters */ atomic_t count; @@ -56,7 +56,7 @@ struct mutex { struct task_struct *owner; #endif #ifdef CONFIG_MUTEX_SPIN_ON_OWNER - struct mcs_spinlock *mcs_lock; /* Spinner MCS lock */ + struct optimistic_spin_queue *osq; /* Spinner MCS lock */ #endif #ifdef CONFIG_DEBUG_MUTEXES const char *name; --- a/kernel/locking/Makefile +++ b/kernel/locking/Makefile @@ -1,5 +1,5 @@ -obj-y += mutex.o semaphore.o rwsem.o lglock.o +obj-y += mutex.o semaphore.o rwsem.o lglock.o mcs_spinlock.o ifdef CONFIG_FUNCTION_TRACER CFLAGS_REMOVE_lockdep.o = -pg --- /dev/null +++ b/kernel/locking/mcs_spinlock.c @@ -0,0 +1,167 @@ + +#include +#include +#include +#include "mcs_spinlock.h" + +#ifdef CONFIG_SMP + +/* + * An MCS like lock especially tailored for optimistic spinning for sleeping + * lock implementations (mutex, rwsem, etc). + * + * Using a single mcs node per CPU is safe because sleeping locks should not be + * called from interrupt context and we have preemption disabled while + * spinning. + */ +static DEFINE_PER_CPU_SHARED_ALIGNED(struct optimistic_spin_queue, osq_node); + +/* + * Get a stable @node->next pointer, either for unlock() or unqueue() purposes. + * Can return NULL in case we were the last queued and we updated @lock instead. + */ +static inline struct optimistic_spin_queue * +osq_wait_next(struct optimistic_spin_queue **lock, + struct optimistic_spin_queue *node, + struct optimistic_spin_queue *prev) +{ + struct optimistic_spin_queue *next = NULL; + + for (;;) { + if (*lock == node && cmpxchg(lock, node, prev) == node) { + /* + * We were the last queued, we moved @lock back. @prev + * will now observe @lock and will complete its + * unlock()/unqueue(). + */ + break; + } + + /* + * We must xchg() the @node->next value, because if we were to + * leave it in, a concurrent unlock()/unqueue() from + * @node->next might complete Step-A and think its @prev is + * still valid. + * + * If the concurrent unlock()/unqueue() wins the race, we'll + * wait for either @lock to point to us, through its Step-B, or + * wait for a new @node->next from its Step-C. + */ + if (node->next) { + next = xchg(&node->next, NULL); + if (next) + break; + } + + arch_mutex_cpu_relax(); + } + + return next; +} + +bool osq_lock(struct optimistic_spin_queue **lock) +{ + struct optimistic_spin_queue *node = this_cpu_ptr(&osq_node); + struct optimistic_spin_queue *prev, *next; + + node->locked = 0; + node->next = NULL; + + node->prev = prev = xchg(lock, node); + if (likely(prev == NULL)) + return true; + + ACCESS_ONCE(prev->next) = node; + + /* + * Normally @prev is untouchable after the above store; because at that + * moment unlock can proceed and wipe the node element from stack. + * + * However, since our nodes are static per-cpu storage, we're + * guaranteed their existence -- this allows us to apply + * cmpxchg in an attempt to undo our queueing. + */ + + while (!smp_load_acquire(&node->locked)) { + /* + * If we need to reschedule bail... so we can block. + */ + if (need_resched()) + goto unqueue; + + arch_mutex_cpu_relax(); + } + return true; + +unqueue: + /* + * Step - A -- stabilize @prev + * + * Undo our @prev->next assignment; this will make @prev's + * unlock()/unqueue() wait for a next pointer since @lock points to us + * (or later). + */ + + for (;;) { + if (prev->next == node && + cmpxchg(&prev->next, node, NULL) == node) + break; + + /* + * We can only fail the cmpxchg() racing against an unlock(), + * in which case we should observe @node->locked becomming + * true. + */ + if (smp_load_acquire(&node->locked)) + return true; + + /* + * Or we race against a concurrent unqueue()'s step-B, in which + * case its step-C will write us a new @node->prev pointer. + */ + prev = ACCESS_ONCE(node->prev); + } + + /* + * Step - B -- stabilize @next + * + * Similar to unlock(), wait for @node->next or move @lock from @node + * back to @prev. + */ + + next = osq_wait_next(lock, node, prev); + if (!next) + return false; + + /* + * Step - C -- unlink + * + * @prev is stable because its still waiting for a new @prev->next + * pointer, @next is stable because our @node->next pointer is NULL and + * it will wait in Step-A. + */ + + ACCESS_ONCE(next->prev) = prev; + ACCESS_ONCE(prev->next) = next; + + return false; +} + +void osq_unlock(struct optimistic_spin_queue **lock) +{ + struct optimistic_spin_queue *node = this_cpu_ptr(&osq_node); + struct optimistic_spin_queue *next; + + /* + * Fast path for the uncontended case. + */ + if (likely(cmpxchg(lock, node, NULL) == node)) + return; + + next = osq_wait_next(lock, node, NULL); + if (next) + ACCESS_ONCE(next->locked) = 1; +} + +#endif + --- a/kernel/locking/mcs_spinlock.h +++ b/kernel/locking/mcs_spinlock.h @@ -111,4 +111,19 @@ void mcs_spin_unlock(struct mcs_spinlock arch_mcs_spin_unlock_contended(&next->locked); } +/* + * Cancellable version of the MCS lock above. + * + * Intended for adaptive spinning of sleeping locks: + * mutex_lock()/rwsem_down_{read,write}() etc. + */ + +struct optimistic_spin_queue { + struct optimistic_spin_queue *next, *prev; + int locked; /* 1 if lock acquired */ +}; + +extern bool osq_lock(struct optimistic_spin_queue **lock); +extern void osq_unlock(struct optimistic_spin_queue **lock); + #endif /* __LINUX_MCS_SPINLOCK_H */ --- a/kernel/locking/mutex.c +++ b/kernel/locking/mutex.c @@ -53,7 +53,7 @@ __mutex_init(struct mutex *lock, const c INIT_LIST_HEAD(&lock->wait_list); mutex_clear_owner(lock); #ifdef CONFIG_MUTEX_SPIN_ON_OWNER - lock->mcs_lock = NULL; + lock->osq = NULL; #endif debug_mutex_init(lock, name, key); @@ -403,7 +403,9 @@ __mutex_lock_common(struct mutex *lock, if (!mutex_can_spin_on_owner(lock)) goto slowpath; - mcs_spin_lock(&lock->mcs_lock); + if (!osq_lock(&lock->osq)) + goto slowpath; + for (;;) { struct task_struct *owner; @@ -442,7 +444,7 @@ __mutex_lock_common(struct mutex *lock, } mutex_set_owner(lock); - mcs_spin_unlock(&lock->mcs_lock); + osq_unlock(&lock->osq); preempt_enable(); return 0; } @@ -464,7 +466,7 @@ __mutex_lock_common(struct mutex *lock, */ arch_mutex_cpu_relax(); } - mcs_spin_unlock(&lock->mcs_lock); + osq_unlock(&lock->osq); slowpath: #endif spin_lock_mutex(&lock->wait_lock, flags); -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/