The Tor Browser / file revision

 /* -*- 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 prthread_h___

 #define prthread_h___

 /*

 ** API for NSPR threads. On some architectures (Mac OS Classic

 ** notably) pre-emptibility is not guaranteed. Hard priority scheduling

 ** is not guaranteed, so programming using priority based synchronization

 ** is a no-no.

 **

 ** NSPR threads are scheduled based loosely on their client set priority.

 ** In general, a thread of a higher priority has a statistically better

 ** chance of running relative to threads of lower priority. However,

 ** NSPR uses multiple strategies to provide execution vehicles for thread

 ** abstraction of various host platforms. As it turns out, there is little

 ** NSPR can do to affect the scheduling attributes of "GLOBAL" threads.

 ** However, a semblance of GLOBAL threads is used to implement "LOCAL"

 ** threads. An arbitrary number of such LOCAL threads can be assigned to

 ** a single GLOBAL thread.

 **

 ** For scheduling, NSPR will attempt to run the highest priority LOCAL

 ** thread associated with a given GLOBAL thread. It is further assumed

 ** that the host OS will apply some form of "fair" scheduling on the

 ** GLOBAL threads.

 **

 ** Threads have a "system flag" which when set indicates the thread

 ** doesn't count for determining when the process should exit (the

 ** process exits when the last user thread exits).

 **

 ** Threads also have a "scope flag" which controls whether the threads

 ** are scheduled in the local scope or scheduled by the OS globally. This 

 ** indicates whether a thread is permanently bound to a native OS thread. 

 ** An unbound thread competes for scheduling resources in the same process.

 **

 ** Another flag is "state flag" which control whether the thread is joinable.

 ** It allows other threads to wait for the created thread to reach completion.

 **

 ** Threads can have "per-thread-data" attached to them. Each thread has a

 ** per-thread error number and error string which are updated when NSPR

 ** operations fail.

 */

 #include "prtypes.h"

 #include "prinrval.h"

 PR_BEGIN_EXTERN_C

 typedef struct PRThread PRThread;

 typedef struct PRThreadStack PRThreadStack;

 typedef enum PRThreadType {

     PR_USER_THREAD,

     PR_SYSTEM_THREAD

 } PRThreadType;

 typedef enum PRThreadScope {

     PR_LOCAL_THREAD,

     PR_GLOBAL_THREAD,

     PR_GLOBAL_BOUND_THREAD

 } PRThreadScope;

 typedef enum PRThreadState {

     PR_JOINABLE_THREAD,

     PR_UNJOINABLE_THREAD

 } PRThreadState;

 typedef enum PRThreadPriority

 {

     PR_PRIORITY_FIRST = 0,      /* just a placeholder */

     PR_PRIORITY_LOW = 0,        /* the lowest possible priority */

     PR_PRIORITY_NORMAL = 1,     /* most common expected priority */

     PR_PRIORITY_HIGH = 2,       /* slightly more aggressive scheduling */

     PR_PRIORITY_URGENT = 3,     /* it does little good to have more than one */

     PR_PRIORITY_LAST = 3        /* this is just a placeholder */

 } PRThreadPriority;

 /*

 ** Create a new thread:

 **     "type" is the type of thread to create

 **     "start(arg)" will be invoked as the threads "main"

 **     "priority" will be created thread's priority

 **     "scope" will specify whether the thread is local or global

 **     "state" will specify whether the thread is joinable or not

 **     "stackSize" the size of the stack, in bytes. The value can be zero

 **        and then a machine specific stack size will be chosen.

 **

 ** This can return NULL if some kind of error occurs, such as if memory is

 ** tight.

 **

 ** If you want the thread to start up waiting for the creator to do

 ** something, enter a lock before creating the thread and then have the

 ** threads start routine enter and exit the same lock. When you are ready

 ** for the thread to run, exit the lock.

 **

 ** If you want to detect the completion of the created thread, the thread

 ** should be created joinable.  Then, use PR_JoinThread to synchrnoize the

 ** termination of another thread.

 **

 ** When the start function returns the thread exits. If it is the last

 ** PR_USER_THREAD to exit then the process exits.

 */

 NSPR_API(PRThread*) PR_CreateThread(PRThreadType type,

                      void (PR_CALLBACK *start)(void *arg),

                      void *arg,

                      PRThreadPriority priority,

                      PRThreadScope scope,

                      PRThreadState state,

                      PRUint32 stackSize);

 /*

 ** Wait for thread termination:

 **     "thread" is the target thread 

 **

 ** This can return PR_FAILURE if no joinable thread could be found 

 ** corresponding to the specified target thread.

 **

 ** The calling thread is blocked until the target thread completes.

 ** Several threads cannot wait for the same thread to complete; one thread

 ** will operate successfully and others will terminate with an error PR_FAILURE.

 ** The calling thread will not be blocked if the target thread has already

 ** terminated.

 */

 NSPR_API(PRStatus) PR_JoinThread(PRThread *thread);

 /*

 ** Return the current thread object for the currently running code.

 ** Never returns NULL.

 */

 NSPR_API(PRThread*) PR_GetCurrentThread(void);

 #ifndef NO_NSPR_10_SUPPORT

 #define PR_CurrentThread() PR_GetCurrentThread() /* for nspr1.0 compat. */

 #endif /* NO_NSPR_10_SUPPORT */

 /*

 ** Get the priority of "thread".

 */

 NSPR_API(PRThreadPriority) PR_GetThreadPriority(const PRThread *thread);

 /*

 ** Change the priority of the "thread" to "priority".

 **

 ** PR_SetThreadPriority works in a best-effort manner. On some platforms a

 ** special privilege, such as root access, is required to change thread

 ** priorities, especially to raise thread priorities. If the caller doesn't

 ** have enough privileges to change thread priorites, the function has no

 ** effect except causing a future PR_GetThreadPriority call to return

 ** |priority|.

 */

 NSPR_API(void) PR_SetThreadPriority(PRThread *thread, PRThreadPriority priority);

 /*

 ** Set the name of the current thread, which will be visible in a debugger

 ** and accessible via a call to PR_GetThreadName().

 */

 NSPR_API(PRStatus) PR_SetCurrentThreadName(const char *name);

 /*

 ** Return the name of "thread", if set.  Otherwise return NULL.

 */

 NSPR_API(const char *) PR_GetThreadName(const PRThread *thread);

 /*

 ** This routine returns a new index for per-thread-private data table. 

 ** The index is visible to all threads within a process. This index can 

 ** be used with the PR_SetThreadPrivate() and PR_GetThreadPrivate() routines 

 ** to save and retrieve data associated with the index for a thread.

 **

 ** Each index is associationed with a destructor function ('dtor'). The function

 ** may be specified as NULL when the index is created. If it is not NULL, the

 ** function will be called when:

 **      - the thread exits and the private data for the associated index

 **        is not NULL,

 **      - new thread private data is set and the current private data is

 **        not NULL.

 **

 ** The index independently maintains specific values for each binding thread. 

 ** A thread can only get access to its own thread-specific-data.

 **

 ** Upon a new index return the value associated with the index for all threads

 ** is NULL, and upon thread creation the value associated with all indices for 

 ** that thread is NULL. 

 **

 ** Returns PR_FAILURE if the total number of indices will exceed the maximun 

 ** allowed.

 */

 typedef void (PR_CALLBACK *PRThreadPrivateDTOR)(void *priv);

 NSPR_API(PRStatus) PR_NewThreadPrivateIndex(

     PRUintn *newIndex, PRThreadPrivateDTOR destructor);

 /*

 ** Define some per-thread-private data.

 **     "tpdIndex" is an index into the per-thread private data table

 **     "priv" is the per-thread-private data 

 **

 ** If the per-thread private data table has a previously registered

 ** destructor function and a non-NULL per-thread-private data value,

 ** the destructor function is invoked.

 **

 ** This can return PR_FAILURE if the index is invalid.

 */

 NSPR_API(PRStatus) PR_SetThreadPrivate(PRUintn tpdIndex, void *priv);

 /*

 ** Recover the per-thread-private data for the current thread. "tpdIndex" is

 ** the index into the per-thread private data table. 

 **

 ** The returned value may be NULL which is indistinguishable from an error 

 ** condition.

 **

 ** A thread can only get access to its own thread-specific-data.

 */

 NSPR_API(void*) PR_GetThreadPrivate(PRUintn tpdIndex);

 /*

 ** This routine sets the interrupt request for a target thread. The interrupt

 ** request remains in the thread's state until it is delivered exactly once

 ** or explicitly canceled.

 **

 ** A thread that has been interrupted will fail all NSPR blocking operations

 ** that return a PRStatus (I/O, waiting on a condition, etc).

 **

 ** PR_Interrupt may itself fail if the target thread is invalid.

 */

 NSPR_API(PRStatus) PR_Interrupt(PRThread *thread);

 /*

 ** Clear the interrupt request for the calling thread. If no such request

 ** is pending, this operation is a noop.

 */

 NSPR_API(void) PR_ClearInterrupt(void);

 /*

 ** Block the interrupt for the calling thread.

 */

 NSPR_API(void) PR_BlockInterrupt(void);

 /*

 ** Unblock the interrupt for the calling thread.

 */

 NSPR_API(void) PR_UnblockInterrupt(void);

 /*

 ** Make the current thread sleep until "ticks" time amount of time

 ** has expired. If "ticks" is PR_INTERVAL_NO_WAIT then the call is

 ** equivalent to calling PR_Yield. Calling PR_Sleep with an argument

 ** equivalent to PR_INTERVAL_NO_TIMEOUT is an error and will result

 ** in a PR_FAILURE error return.

 */

 NSPR_API(PRStatus) PR_Sleep(PRIntervalTime ticks);

 /*

 ** Get the scoping of this thread.

 */

 NSPR_API(PRThreadScope) PR_GetThreadScope(const PRThread *thread);

 /*

 ** Get the type of this thread.

 */

 NSPR_API(PRThreadType) PR_GetThreadType(const PRThread *thread);

 /*

 ** Get the join state of this thread.

 */

 NSPR_API(PRThreadState) PR_GetThreadState(const PRThread *thread);

 PR_END_EXTERN_C

 #endif /* prthread_h___ */