2 {-# OPTIONS -fno-implicit-prelude #-}
3 -----------------------------------------------------------------------------
6 -- Copyright : (c) The University of Glasgow, 1994-2002
7 -- License : see libraries/base/LICENSE
9 -- Maintainer : cvs-ghc@haskell.org
10 -- Stability : internal
11 -- Portability : non-portable (GHC extensions)
13 -- Basic concurrency stuff.
15 -----------------------------------------------------------------------------
17 #include "ghcconfig.h"
21 -- Forking and suchlike
22 , myThreadId -- :: IO ThreadId
23 , killThread -- :: ThreadId -> IO ()
24 , throwTo -- :: ThreadId -> Exception -> IO ()
25 , par -- :: a -> b -> b
26 , pseq -- :: a -> b -> b
28 , labelThread -- :: ThreadId -> String -> IO ()
31 , threadDelay -- :: Int -> IO ()
32 , threadWaitRead -- :: Int -> IO ()
33 , threadWaitWrite -- :: Int -> IO ()
37 , newMVar -- :: a -> IO (MVar a)
38 , newEmptyMVar -- :: IO (MVar a)
39 , takeMVar -- :: MVar a -> IO a
40 , putMVar -- :: MVar a -> a -> IO ()
41 , tryTakeMVar -- :: MVar a -> IO (Maybe a)
42 , tryPutMVar -- :: MVar a -> a -> IO Bool
43 , isEmptyMVar -- :: MVar a -> IO Bool
44 , addMVarFinalizer -- :: MVar a -> IO () -> IO ()
48 , atomically -- :: STM a -> IO a
50 , orElse -- :: STM a -> STM a -> STM a
51 , catchSTM -- :: STM a -> (Exception -> STM a) -> STM a
53 , newTVar -- :: a -> STM (TVar a)
54 , readTVar -- :: TVar a -> STM a
55 , writeTVar -- :: a -> TVar a -> STM ()
57 #ifdef mingw32_TARGET_OS
58 , asyncRead -- :: Int -> Int -> Int -> Ptr a -> IO (Int, Int)
59 , asyncWrite -- :: Int -> Int -> Int -> Ptr a -> IO (Int, Int)
60 , asyncDoProc -- :: FunPtr (Ptr a -> IO Int) -> Ptr a -> IO Int
62 , asyncReadBA -- :: Int -> Int -> Int -> Int -> MutableByteArray# RealWorld -> IO (Int, Int)
63 , asyncWriteBA -- :: Int -> Int -> Int -> Int -> MutableByteArray# RealWorld -> IO (Int, Int)
67 import System.Posix.Types
68 import System.Posix.Internals
76 import GHC.Num ( Num(..) )
77 import GHC.Real ( fromIntegral, quot )
78 import GHC.Base ( Int(..) )
79 import GHC.Exception ( Exception(..), AsyncException(..) )
80 import GHC.Pack ( packCString# )
81 import GHC.Ptr ( Ptr(..), plusPtr, FunPtr(..) )
84 infixr 0 `par`, `pseq`
87 %************************************************************************
89 \subsection{@ThreadId@, @par@, and @fork@}
91 %************************************************************************
94 data ThreadId = ThreadId ThreadId#
95 -- ToDo: data ThreadId = ThreadId (Weak ThreadId#)
96 -- But since ThreadId# is unlifted, the Weak type must use open
99 A 'ThreadId' is an abstract type representing a handle to a thread.
100 'ThreadId' is an instance of 'Eq', 'Ord' and 'Show', where
101 the 'Ord' instance implements an arbitrary total ordering over
102 'ThreadId's. The 'Show' instance lets you convert an arbitrary-valued
103 'ThreadId' to string form; showing a 'ThreadId' value is occasionally
104 useful when debugging or diagnosing the behaviour of a concurrent
107 /Note/: in GHC, if you have a 'ThreadId', you essentially have
108 a pointer to the thread itself. This means the thread itself can\'t be
109 garbage collected until you drop the 'ThreadId'.
110 This misfeature will hopefully be corrected at a later date.
112 /Note/: Hugs does not provide any operations on other threads;
113 it defines 'ThreadId' as a synonym for ().
116 --forkIO has now been hoisted out into the Concurrent library.
118 {- | 'killThread' terminates the given thread (GHC only).
119 Any work already done by the thread isn\'t
120 lost: the computation is suspended until required by another thread.
121 The memory used by the thread will be garbage collected if it isn\'t
122 referenced from anywhere. The 'killThread' function is defined in
125 > killThread tid = throwTo tid (AsyncException ThreadKilled)
128 killThread :: ThreadId -> IO ()
129 killThread tid = throwTo tid (AsyncException ThreadKilled)
131 {- | 'throwTo' raises an arbitrary exception in the target thread (GHC only).
133 'throwTo' does not return until the exception has been raised in the
134 target thread. The calling thread can thus be certain that the target
135 thread has received the exception. This is a useful property to know
136 when dealing with race conditions: eg. if there are two threads that
137 can kill each other, it is guaranteed that only one of the threads
138 will get to kill the other. -}
139 throwTo :: ThreadId -> Exception -> IO ()
140 throwTo (ThreadId id) ex = IO $ \ s ->
141 case (killThread# id ex s) of s1 -> (# s1, () #)
143 -- | Returns the 'ThreadId' of the calling thread (GHC only).
144 myThreadId :: IO ThreadId
145 myThreadId = IO $ \s ->
146 case (myThreadId# s) of (# s1, id #) -> (# s1, ThreadId id #)
149 -- |The 'yield' action allows (forces, in a co-operative multitasking
150 -- implementation) a context-switch to any other currently runnable
151 -- threads (if any), and is occasionally useful when implementing
152 -- concurrency abstractions.
155 case (yield# s) of s1 -> (# s1, () #)
157 {- | 'labelThread' stores a string as identifier for this thread if
158 you built a RTS with debugging support. This identifier will be used in
159 the debugging output to make distinction of different threads easier
160 (otherwise you only have the thread state object\'s address in the heap).
162 Other applications like the graphical Concurrent Haskell Debugger
163 (<http://www.informatik.uni-kiel.de/~fhu/chd/>) may choose to overload
164 'labelThread' for their purposes as well.
167 labelThread :: ThreadId -> String -> IO ()
168 labelThread (ThreadId t) str = IO $ \ s ->
169 let ps = packCString# str
170 adr = byteArrayContents# ps in
171 case (labelThread# t adr s) of s1 -> (# s1, () #)
173 -- Nota Bene: 'pseq' used to be 'seq'
174 -- but 'seq' is now defined in PrelGHC
176 -- "pseq" is defined a bit weirdly (see below)
178 -- The reason for the strange "lazy" call is that
179 -- it fools the compiler into thinking that pseq and par are non-strict in
180 -- their second argument (even if it inlines pseq at the call site).
181 -- If it thinks pseq is strict in "y", then it often evaluates
182 -- "y" before "x", which is totally wrong.
186 pseq x y = x `seq` lazy y
190 par x y = case (par# x) of { _ -> lazy y }
194 %************************************************************************
196 \subsection[stm]{Transactional heap operations}
198 %************************************************************************
200 TVars are shared memory locations which support atomic memory
204 newtype STM a = STM (State# RealWorld -> (# State# RealWorld, a #))
206 unSTM :: STM a -> (State# RealWorld -> (# State# RealWorld, a #))
209 instance Functor STM where
210 fmap f x = x >>= (return . f)
212 instance Monad STM where
213 {-# INLINE return #-}
216 m >> k = m >>= \_ -> k
217 return x = returnSTM x
218 m >>= k = bindSTM m k
220 bindSTM :: STM a -> (a -> STM b) -> STM b
221 bindSTM (STM m) k = STM ( \s ->
223 (# new_s, a #) -> unSTM (k a) new_s
226 thenSTM :: STM a -> STM b -> STM b
227 thenSTM (STM m) k = STM ( \s ->
229 (# new_s, a #) -> unSTM k new_s
232 returnSTM :: a -> STM a
233 returnSTM x = STM (\s -> (# s, x #))
235 -- |Perform a series of STM actions atomically.
236 atomically :: STM a -> IO a
237 atomically (STM m) = IO (\s -> (atomically# m) s )
239 -- |Retry execution of the current memory transaction because it has seen
240 -- values in TVars which mean that it should not continue (e.g. the TVars
241 -- represent a shared buffer that is now empty). The implementation may
242 -- block the thread until one of the TVars that it has read from has been
245 retry = STM $ \s# -> retry# s#
247 -- |Compose two alternative STM actions. If the first action completes without
248 -- retrying then it forms the result of the orElse. Otherwise, if the first
249 -- action retries, then the second action is tried in its place. If both actions
250 -- retry then the orElse as a whole retries.
251 orElse :: STM a -> STM a -> STM a
252 orElse (STM m) e = STM $ \s -> catchRetry# m (unSTM e) s
254 -- |Exception handling within STM actions.
255 catchSTM :: STM a -> (Exception -> STM a) -> STM a
256 catchSTM (STM m) k = STM $ \s -> catchSTM# m (\ex -> unSTM (k ex)) s
258 data TVar a = TVar (TVar# RealWorld a)
260 instance Eq (TVar a) where
261 (TVar tvar1#) == (TVar tvar2#) = sameTVar# tvar1# tvar2#
263 -- |Create a new TVar holding a value supplied
264 newTVar :: a -> STM (TVar a)
265 newTVar val = STM $ \s1# ->
266 case newTVar# val s1# of
267 (# s2#, tvar# #) -> (# s2#, TVar tvar# #)
269 -- |Return the current value stored in a TVar
270 readTVar :: TVar a -> STM a
271 readTVar (TVar tvar#) = STM $ \s# -> readTVar# tvar# s#
273 -- |Write the supplied value into a TVar
274 writeTVar :: TVar a -> a -> STM ()
275 writeTVar (TVar tvar#) val = STM $ \s1# ->
276 case writeTVar# tvar# val s1# of
281 %************************************************************************
283 \subsection[mvars]{M-Structures}
285 %************************************************************************
287 M-Vars are rendezvous points for concurrent threads. They begin
288 empty, and any attempt to read an empty M-Var blocks. When an M-Var
289 is written, a single blocked thread may be freed. Reading an M-Var
290 toggles its state from full back to empty. Therefore, any value
291 written to an M-Var may only be read once. Multiple reads and writes
292 are allowed, but there must be at least one read between any two
296 --Defined in IOBase to avoid cycle: data MVar a = MVar (SynchVar# RealWorld a)
298 -- |Create an 'MVar' which is initially empty.
299 newEmptyMVar :: IO (MVar a)
300 newEmptyMVar = IO $ \ s# ->
302 (# s2#, svar# #) -> (# s2#, MVar svar# #)
304 -- |Create an 'MVar' which contains the supplied value.
305 newMVar :: a -> IO (MVar a)
307 newEmptyMVar >>= \ mvar ->
308 putMVar mvar value >>
311 -- |Return the contents of the 'MVar'. If the 'MVar' is currently
312 -- empty, 'takeMVar' will wait until it is full. After a 'takeMVar',
313 -- the 'MVar' is left empty.
315 -- If several threads are competing to take the same 'MVar', one is chosen
316 -- to continue at random when the 'MVar' becomes full.
317 takeMVar :: MVar a -> IO a
318 takeMVar (MVar mvar#) = IO $ \ s# -> takeMVar# mvar# s#
320 -- |Put a value into an 'MVar'. If the 'MVar' is currently full,
321 -- 'putMVar' will wait until it becomes empty.
323 -- If several threads are competing to fill the same 'MVar', one is
324 -- chosen to continue at random when the 'MVar' becomes empty.
325 putMVar :: MVar a -> a -> IO ()
326 putMVar (MVar mvar#) x = IO $ \ s# ->
327 case putMVar# mvar# x s# of
330 -- |A non-blocking version of 'takeMVar'. The 'tryTakeMVar' function
331 -- returns immediately, with 'Nothing' if the 'MVar' was empty, or
332 -- @'Just' a@ if the 'MVar' was full with contents @a@. After 'tryTakeMVar',
333 -- the 'MVar' is left empty.
334 tryTakeMVar :: MVar a -> IO (Maybe a)
335 tryTakeMVar (MVar m) = IO $ \ s ->
336 case tryTakeMVar# m s of
337 (# s, 0#, _ #) -> (# s, Nothing #) -- MVar is empty
338 (# s, _, a #) -> (# s, Just a #) -- MVar is full
340 -- |A non-blocking version of 'putMVar'. The 'tryPutMVar' function
341 -- attempts to put the value @a@ into the 'MVar', returning 'True' if
342 -- it was successful, or 'False' otherwise.
343 tryPutMVar :: MVar a -> a -> IO Bool
344 tryPutMVar (MVar mvar#) x = IO $ \ s# ->
345 case tryPutMVar# mvar# x s# of
346 (# s, 0# #) -> (# s, False #)
347 (# s, _ #) -> (# s, True #)
349 -- |Check whether a given 'MVar' is empty.
351 -- Notice that the boolean value returned is just a snapshot of
352 -- the state of the MVar. By the time you get to react on its result,
353 -- the MVar may have been filled (or emptied) - so be extremely
354 -- careful when using this operation. Use 'tryTakeMVar' instead if possible.
355 isEmptyMVar :: MVar a -> IO Bool
356 isEmptyMVar (MVar mv#) = IO $ \ s# ->
357 case isEmptyMVar# mv# s# of
358 (# s2#, flg #) -> (# s2#, not (flg ==# 0#) #)
360 -- |Add a finalizer to an 'MVar' (GHC only). See "Foreign.ForeignPtr" and
361 -- "System.Mem.Weak" for more about finalizers.
362 addMVarFinalizer :: MVar a -> IO () -> IO ()
363 addMVarFinalizer (MVar m) finalizer =
364 IO $ \s -> case mkWeak# m () finalizer s of { (# s1, w #) -> (# s1, () #) }
368 %************************************************************************
370 \subsection{Thread waiting}
372 %************************************************************************
375 #ifdef mingw32_TARGET_OS
377 -- Note: threadDelay, threadWaitRead and threadWaitWrite aren't really functional
378 -- on Win32, but left in there because lib code (still) uses them (the manner
379 -- in which they're used doesn't cause problems on a Win32 platform though.)
381 asyncRead :: Int -> Int -> Int -> Ptr a -> IO (Int, Int)
382 asyncRead (I# fd) (I# isSock) (I# len) (Ptr buf) = do
383 (l, rc) <- IO (\s -> case asyncRead# fd isSock len buf s of
384 (# s, len#, err# #) -> (# s, (I# len#, I# err#) #))
385 -- special handling for Ctrl+C-aborted 'standard input' reads;
386 -- see rts/win32/ConsoleHandler.c for details.
387 if (l == 0 && rc == -2)
388 then asyncRead (I# fd) (I# isSock) (I# len) (Ptr buf)
391 asyncWrite :: Int -> Int -> Int -> Ptr a -> IO (Int, Int)
392 asyncWrite (I# fd) (I# isSock) (I# len) (Ptr buf) =
393 IO $ \s -> case asyncWrite# fd isSock len buf s of
394 (# s, len#, err# #) -> (# s, (I# len#, I# err#) #)
396 asyncDoProc :: FunPtr (Ptr a -> IO Int) -> Ptr a -> IO Int
397 asyncDoProc (FunPtr proc) (Ptr param) =
398 -- the 'length' value is ignored; simplifies implementation of
399 -- the async*# primops to have them all return the same result.
400 IO $ \s -> case asyncDoProc# proc param s of
401 (# s, len#, err# #) -> (# s, I# err# #)
403 -- to aid the use of these primops by the IO Handle implementation,
404 -- provide the following convenience funs:
406 -- this better be a pinned byte array!
407 asyncReadBA :: Int -> Int -> Int -> Int -> MutableByteArray# RealWorld -> IO (Int,Int)
408 asyncReadBA fd isSock len off bufB =
409 asyncRead fd isSock len ((Ptr (byteArrayContents# (unsafeCoerce# bufB))) `plusPtr` off)
411 asyncWriteBA :: Int -> Int -> Int -> Int -> MutableByteArray# RealWorld -> IO (Int,Int)
412 asyncWriteBA fd isSock len off bufB =
413 asyncWrite fd isSock len ((Ptr (byteArrayContents# (unsafeCoerce# bufB))) `plusPtr` off)
417 -- -----------------------------------------------------------------------------
420 -- | Block the current thread until data is available to read on the
421 -- given file descriptor (GHC only).
422 threadWaitRead :: Fd -> IO ()
424 #ifndef mingw32_TARGET_OS
425 | threaded = waitForReadEvent fd
427 | otherwise = IO $ \s ->
428 case fromIntegral fd of { I# fd# ->
429 case waitRead# fd# s of { s -> (# s, () #)
432 -- | Block the current thread until data can be written to the
433 -- given file descriptor (GHC only).
434 threadWaitWrite :: Fd -> IO ()
436 #ifndef mingw32_TARGET_OS
437 | threaded = waitForWriteEvent fd
439 | otherwise = IO $ \s ->
440 case fromIntegral fd of { I# fd# ->
441 case waitWrite# fd# s of { s -> (# s, () #)
444 -- | Suspends the current thread for a given number of microseconds
447 -- Note that the resolution used by the Haskell runtime system's
448 -- internal timer is 1\/50 second, and 'threadDelay' will round its
449 -- argument up to the nearest multiple of this resolution.
451 -- There is no guarantee that the thread will be rescheduled promptly
452 -- when the delay has expired, but the thread will never continue to
453 -- run /earlier/ than specified.
455 threadDelay :: Int -> IO ()
457 #ifndef mingw32_TARGET_OS
458 | threaded = waitForDelayEvent time
460 | threaded = c_Sleep (fromIntegral (time `quot` 1000))
462 | otherwise = IO $ \s ->
463 case fromIntegral time of { I# time# ->
464 case delay# time# s of { s -> (# s, () #)
467 -- On Windows, we just make a safe call to 'Sleep' to implement threadDelay.
468 #ifdef mingw32_TARGET_OS
469 foreign import ccall safe "Sleep" c_Sleep :: CInt -> IO ()
472 foreign import ccall unsafe "rtsSupportsBoundThreads" threaded :: Bool
474 -- ----------------------------------------------------------------------------
475 -- Threaded RTS implementation of threadWaitRead, threadWaitWrite, threadDelay
477 -- In the threaded RTS, we employ a single IO Manager thread to wait
478 -- for all outstanding IO requests (threadWaitRead,threadWaitWrite)
479 -- and delays (threadDelay).
481 -- We can do this because in the threaded RTS the IO Manager can make
482 -- a non-blocking call to select(), so we don't have to do select() in
483 -- the scheduler as we have to in the non-threaded RTS. We get performance
484 -- benefits from doing it this way, because we only have to restart the select()
485 -- when a new request arrives, rather than doing one select() each time
486 -- around the scheduler loop. Furthermore, the scheduler can be simplified
487 -- by not having to check for completed IO requests.
489 -- Issues, possible problems:
491 -- - we might want bound threads to just do the blocking
492 -- operation rather than communicating with the IO manager
493 -- thread. This would prevent simgle-threaded programs which do
494 -- IO from requiring multiple OS threads. However, it would also
495 -- prevent bound threads waiting on IO from being killed or sent
498 -- - Apprently exec() doesn't work on Linux in a multithreaded program.
499 -- I couldn't repeat this.
501 -- - How do we handle signal delivery in the multithreaded RTS?
503 -- - forkProcess will kill the IO manager thread. Let's just
504 -- hope we don't need to do any blocking IO between fork & exec.
506 #ifndef mingw32_TARGET_OS
509 = Read {-# UNPACK #-} !Fd {-# UNPACK #-} !(MVar ())
510 | Write {-# UNPACK #-} !Fd {-# UNPACK #-} !(MVar ())
513 = Delay {-# UNPACK #-} !Int {-# UNPACK #-} !(MVar ())
515 pendingEvents :: IORef [IOReq]
516 pendingDelays :: IORef [DelayReq]
517 -- could use a strict list or array here
518 {-# NOINLINE pendingEvents #-}
519 {-# NOINLINE pendingDelays #-}
520 (pendingEvents,pendingDelays) = unsafePerformIO $ do
525 -- the first time we schedule an IO request, the service thread
526 -- will be created (cool, huh?)
528 startIOServiceThread :: IO ()
529 startIOServiceThread = do
530 allocaArray 2 $ \fds -> do
531 throwErrnoIfMinus1 "startIOServiceThread" (c_pipe fds)
532 rd_end <- peekElemOff fds 0
533 wr_end <- peekElemOff fds 1
534 writeIORef stick (fromIntegral wr_end)
536 allocaBytes sizeofFdSet $ \readfds -> do
537 allocaBytes sizeofFdSet $ \writefds -> do
538 allocaBytes sizeofTimeVal $ \timeval -> do
539 service_loop (fromIntegral rd_end) readfds writefds timeval [] []
542 -- XXX: move real forkIO here from Control.Concurrent?
543 quickForkIO action = IO $ \s ->
544 case (fork# action s) of (# s1, id #) -> (# s1, ThreadId id #)
547 :: Fd -- listen to this for wakeup calls
554 service_loop wakeup readfds writefds ptimeval old_reqs old_delays = do
556 -- pick up new IO requests
557 new_reqs <- atomicModifyIORef pendingEvents (\a -> ([],a))
558 let reqs = new_reqs ++ old_reqs
560 -- pick up new delay requests
561 new_delays <- atomicModifyIORef pendingDelays (\a -> ([],a))
562 let delays = foldr insertDelay old_delays new_delays
564 -- build the FDSets for select()
568 maxfd <- buildFdSets 0 readfds writefds reqs
570 -- check the current time and wake up any thread in threadDelay whose
571 -- timeout has expired. Also find the timeout value for the select() call.
573 (delays', timeout) <- getDelay now ptimeval delays
575 -- perform the select()
577 res <- c_select ((max wakeup maxfd)+1) readfds writefds
588 -- ToDo: check result
590 b <- takeMVar prodding
591 if b then alloca $ \p -> do c_read (fromIntegral wakeup) p 1; return ()
593 putMVar prodding False
595 reqs' <- completeRequests reqs readfds writefds []
596 service_loop wakeup readfds writefds ptimeval reqs' delays'
599 {-# NOINLINE stick #-}
600 stick = unsafePerformIO (newIORef 0)
602 prodding :: MVar Bool
603 {-# NOINLINE prodding #-}
604 prodding = unsafePerformIO (newMVar False)
606 prodServiceThread :: IO ()
607 prodServiceThread = do
608 b <- takeMVar prodding
610 then do fd <- readIORef stick
611 with 42 $ \pbuf -> do c_write (fromIntegral fd) pbuf 1; return ()
613 putMVar prodding True
615 -- -----------------------------------------------------------------------------
618 buildFdSets maxfd readfds writefds [] = return maxfd
619 buildFdSets maxfd readfds writefds (Read fd m : reqs) = do
621 buildFdSets (max maxfd fd) readfds writefds reqs
622 buildFdSets maxfd readfds writefds (Write fd m : reqs) = do
624 buildFdSets (max maxfd fd) readfds writefds reqs
626 completeRequests [] _ _ reqs' = return reqs'
627 completeRequests (Read fd m : reqs) readfds writefds reqs' = do
628 b <- fdIsSet fd readfds
630 then do putMVar m (); completeRequests reqs readfds writefds reqs'
631 else completeRequests reqs readfds writefds (Read fd m : reqs')
632 completeRequests (Write fd m : reqs) readfds writefds reqs' = do
633 b <- fdIsSet fd writefds
635 then do putMVar m (); completeRequests reqs readfds writefds reqs'
636 else completeRequests reqs readfds writefds (Write fd m : reqs')
638 waitForReadEvent :: Fd -> IO ()
639 waitForReadEvent fd = do
641 atomicModifyIORef pendingEvents (\xs -> (Read fd m : xs, ()))
645 waitForWriteEvent :: Fd -> IO ()
646 waitForWriteEvent fd = do
648 atomicModifyIORef pendingEvents (\xs -> (Write fd m : xs, ()))
652 -- XXX: move into GHC.IOBase from Data.IORef?
653 atomicModifyIORef :: IORef a -> (a -> (a,b)) -> IO b
654 atomicModifyIORef (IORef (STRef r#)) f = IO $ \s -> atomicModifyMutVar# r# f s
656 -- -----------------------------------------------------------------------------
659 waitForDelayEvent :: Int -> IO ()
660 waitForDelayEvent usecs = do
663 let target = now + usecs `quot` tick_usecs
664 atomicModifyIORef pendingDelays (\xs -> (Delay target m : xs, ()))
668 -- Walk the queue of pending delays, waking up any that have passed
669 -- and return the smallest delay to wait for. The queue of pending
670 -- delays is kept ordered.
671 getDelay :: Ticks -> Ptr CTimeVal -> [DelayReq] -> IO ([DelayReq], Ptr CTimeVal)
672 getDelay now ptimeval [] = return ([],nullPtr)
673 getDelay now ptimeval all@(Delay time m : rest)
676 getDelay now ptimeval rest
678 setTimevalTicks ptimeval (time - now)
679 return (all,ptimeval)
681 insertDelay :: DelayReq -> [DelayReq] -> [DelayReq]
682 insertDelay d@(Delay time m) [] = [d]
683 insertDelay d1@(Delay time m) ds@(d2@(Delay time' m') : rest)
684 | time <= time' = d1 : ds
685 | otherwise = d2 : insertDelay d1 rest
688 tick_freq = 50 :: Ticks -- accuracy of threadDelay (ticks per sec)
689 tick_usecs = 1000000 `quot` tick_freq :: Int
691 newtype CTimeVal = CTimeVal ()
693 foreign import ccall unsafe "sizeofTimeVal"
696 foreign import ccall unsafe "getTicksOfDay"
697 getTicksOfDay :: IO Ticks
699 foreign import ccall unsafe "setTimevalTicks"
700 setTimevalTicks :: Ptr CTimeVal -> Ticks -> IO ()
702 -- ----------------------------------------------------------------------------
703 -- select() interface
705 -- ToDo: move to System.Posix.Internals?
707 newtype CFdSet = CFdSet ()
709 foreign import ccall safe "select"
710 c_select :: Fd -> Ptr CFdSet -> Ptr CFdSet -> Ptr CFdSet -> Ptr CTimeVal
713 foreign import ccall unsafe "hsFD_CLR"
714 fdClr :: Fd -> Ptr CFdSet -> IO ()
716 foreign import ccall unsafe "hsFD_ISSET"
717 fdIsSet :: Fd -> Ptr CFdSet -> IO CInt
719 foreign import ccall unsafe "hsFD_SET"
720 fdSet :: Fd -> Ptr CFdSet -> IO ()
722 foreign import ccall unsafe "hsFD_ZERO"
723 fdZero :: Ptr CFdSet -> IO ()
725 foreign import ccall unsafe "sizeof_fd_set"