gecko/nsprpub/pr/include/prcvar.h
Michael Wu ff73a9422b Update NSPR to NSPR_4_9_1_BETA1 tag, a=ted
Includes bug 331299, 722125, 693329, 274013, 716564, 736962, 739469, 482002, 693329
2012-05-10 12:01:43 -07:00

95 lines
3.5 KiB
C

/* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#ifndef prcvar_h___
#define prcvar_h___
#include "prlock.h"
#include "prinrval.h"
PR_BEGIN_EXTERN_C
typedef struct PRCondVar PRCondVar;
/*
** Create a new condition variable.
**
** "lock" is the lock used to protect the condition variable.
**
** Condition variables are synchronization objects that threads can use
** to wait for some condition to occur.
**
** This may fail if memory is tight or if some operating system resource
** is low. In such cases, a NULL will be returned.
*/
NSPR_API(PRCondVar*) PR_NewCondVar(PRLock *lock);
/*
** Destroy a condition variable. There must be no thread
** waiting on the condvar. The caller is responsible for guaranteeing
** that the condvar is no longer in use.
**
*/
NSPR_API(void) PR_DestroyCondVar(PRCondVar *cvar);
/*
** The thread that waits on a condition is blocked in a "waiting on
** condition" state until another thread notifies the condition or a
** caller specified amount of time expires. The lock associated with
** the condition variable will be released, which must have be held
** prior to the call to wait.
**
** Logically a notified thread is moved from the "waiting on condition"
** state and made "ready." When scheduled, it will attempt to reacquire
** the lock that it held when wait was called.
**
** The timeout has two well known values, PR_INTERVAL_NO_TIMEOUT and
** PR_INTERVAL_NO_WAIT. The former value requires that a condition be
** notified (or the thread interrupted) before it will resume from the
** wait. If the timeout has a value of PR_INTERVAL_NO_WAIT, the effect
** is to release the lock, possibly causing a rescheduling within the
** runtime, then immediately attempting to reacquire the lock and resume.
**
** Any other value for timeout will cause the thread to be rescheduled
** either due to explicit notification or an expired interval. The latter
** must be determined by treating time as one part of the monitored data
** being protected by the lock and tested explicitly for an expired
** interval.
**
** Returns PR_FAILURE if the caller has not locked the lock associated
** with the condition variable or the thread was interrupted (PR_Interrupt()).
** The particular reason can be extracted with PR_GetError().
*/
NSPR_API(PRStatus) PR_WaitCondVar(PRCondVar *cvar, PRIntervalTime timeout);
/*
** Notify ONE thread that is currently waiting on 'cvar'. Which thread is
** dependent on the implementation of the runtime. Common sense would dictate
** that all threads waiting on a single condition have identical semantics,
** therefore which one gets notified is not significant.
**
** The calling thead must hold the lock that protects the condition, as
** well as the invariants that are tightly bound to the condition, when
** notify is called.
**
** Returns PR_FAILURE if the caller has not locked the lock associated
** with the condition variable.
*/
NSPR_API(PRStatus) PR_NotifyCondVar(PRCondVar *cvar);
/*
** Notify all of the threads waiting on the condition variable. The order
** that the threads are notified is indeterminant. The lock that protects
** the condition must be held.
**
** Returns PR_FAILURE if the caller has not locked the lock associated
** with the condition variable.
*/
NSPR_API(PRStatus) PR_NotifyAllCondVar(PRCondVar *cvar);
PR_END_EXTERN_C
#endif /* prcvar_h___ */