[project @ 2003-06-26 12:22:59 by stolz]
[ghc-hetmet.git] / ghc / rts / Select.c
index 1666b49..677fdd2 100644 (file)
@@ -1,23 +1,26 @@
 /* -----------------------------------------------------------------------------
- * $Id: Select.c,v 1.10 2000/03/20 09:42:50 andy Exp $
+ * $Id: Select.c,v 1.29 2003/06/26 12:22:59 stolz Exp $
  *
- * (c) The GHC Team 1995-1999
+ * (c) The GHC Team 1995-2002
  *
  * Support for concurrent non-blocking I/O and thread waiting.
  *
  * ---------------------------------------------------------------------------*/
 
+
 /* we're outside the realms of POSIX here... */
-#define NON_POSIX_SOURCE
+/* #include "PosixSource.h" */
 
 #include "Rts.h"
 #include "Schedule.h"
 #include "RtsUtils.h"
 #include "RtsFlags.h"
+#include "Timer.h"
 #include "Itimer.h"
 #include "Signals.h"
+#include "Capability.h"
 
-# if defined(HAVE_SYS_TYPES_H)
+# ifdef HAVE_SYS_TYPES_H
 #  include <sys/types.h>
 # endif
 
 #  include <sys/time.h>
 # endif
 
-nat ticks_since_select = 0;
+#include <errno.h>
+#include <string.h>
+
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+
+/* last timestamp */
+nat timestamp = 0;
+
+#ifdef RTS_SUPPORTS_THREADS
+static rtsBool isWorkerBlockedInAwaitEvent = rtsFalse;
+static rtsBool workerWakeupPending = rtsFalse;
+static int workerWakeupPipe[2];
+static rtsBool workerWakeupInited = rtsFalse;
+#endif
+
+/* There's a clever trick here to avoid problems when the time wraps
+ * around.  Since our maximum delay is smaller than 31 bits of ticks
+ * (it's actually 31 bits of microseconds), we can safely check
+ * whether a timer has expired even if our timer will wrap around
+ * before the target is reached, using the following formula:
+ *
+ *        (int)((uint)current_time - (uint)target_time) < 0
+ *
+ * if this is true, then our time has expired.
+ * (idea due to Andy Gill).
+ */
+rtsBool
+wakeUpSleepingThreads(nat ticks)
+{
+    StgTSO *tso;
+    rtsBool flag = rtsFalse;
+
+    while (sleeping_queue != END_TSO_QUEUE &&
+          (int)(ticks - sleeping_queue->block_info.target) > 0) {
+       tso = sleeping_queue;
+       sleeping_queue = tso->link;
+       tso->why_blocked = NotBlocked;
+       tso->link = END_TSO_QUEUE;
+       IF_DEBUG(scheduler,belch("Waking up sleeping thread %d\n", tso->id));
+       PUSH_ON_RUN_QUEUE(tso);
+       flag = rtsTrue;
+    }
+    return flag;
+}
 
 /* Argument 'wait' says whether to wait for I/O to become available,
  * or whether to just check and return immediately.  If there are
@@ -33,32 +81,36 @@ nat ticks_since_select = 0;
  * otherwise we wait (see Schedule.c).
  *
  * SMP note: must be called with sched_mutex locked.
+ *
+ * Windows: select only works on sockets, so this doesn't really work,
+ * though it makes things better than before. MsgWaitForMultipleObjects
+ * should really be used, though it only seems to work for read handles,
+ * not write handles.
+ *
  */
 void
 awaitEvent(rtsBool wait)
 {
-#ifdef mingw32_TARGET_OS
-/*
- * Win32 doesn't support select(). ToDo: use MsgWaitForMultipleObjects()
- * to achieve (similar) effect.
- *
- */
-    return;
-#else
-
     StgTSO *tso, *prev, *next;
     rtsBool ready;
     fd_set rfd,wfd;
     int numFound;
-    nat min, delta;
     int maxfd = -1;
-   
+    rtsBool select_succeeded = rtsTrue;
+    rtsBool unblock_all = rtsFalse;
     struct timeval tv;
-#ifndef linux_TARGET_OS
-    struct timeval tv_before,tv_after;
-#endif
+    lnat min, ticks;
 
-    IF_DEBUG(scheduler,belch("Checking for threads blocked on I/O...\n"));
+    tv.tv_sec  = 0;
+    tv.tv_usec = 0;
+    
+    IF_DEBUG(scheduler,
+            belch("scheduler: checking for threads blocked on I/O");
+            if (wait) {
+                belch(" (waiting)");
+            }
+            belch("\n");
+            );
 
     /* loop until we've woken up some threads.  This loop is needed
      * because the select timing isn't accurate, we sometimes sleep
@@ -67,18 +119,22 @@ awaitEvent(rtsBool wait)
      */
     do {
 
-      /* see how long it's been since we last checked the blocked queue.
-       * ToDo: make this check atomic, so we don't lose any ticks.
-       */
-      delta = ticks_since_select;
-      ticks_since_select = 0;
-      delta = delta * TICK_MILLISECS * 1000;
+      ticks = timestamp = getourtimeofday();
+      if (wakeUpSleepingThreads(ticks)) { 
+         return;
+      }
 
-      min = wait == rtsTrue ? 0x7fffffff : 0;
+      if (!wait) {
+         min = 0;
+      } else if (sleeping_queue != END_TSO_QUEUE) {
+         min = (sleeping_queue->block_info.target - ticks) 
+             * TICK_MILLISECS * 1000;
+      } else {
+         min = 0x7ffffff;
+      }
 
       /* 
-       * Collect all of the fd's that we're interested in, and capture
-       * the minimum waiting time (in microseconds) for the delayed threads.
+       * Collect all of the fd's that we're interested in
        */
       FD_ZERO(&rfd);
       FD_ZERO(&wfd);
@@ -103,28 +159,20 @@ awaitEvent(rtsBool wait)
            continue;
          }
 
-       case BlockedOnDelay:
-         {
-           int candidate; /* signed int is intentional */
-#if defined(HAVE_SETITIMER)
-           candidate = tso->block_info.delay;
-#else
-           candidate = tso->block_info.target - getourtimeofday();
-           if (candidate < 0) {
-             candidate = 0;
-           }
-#endif
-           if ((nat)candidate < min) {
-             min = candidate;
-           }
-           continue;
-         }
-
        default:
          barf("AwaitEvent");
        }
       }
 
+#ifdef RTS_SUPPORTS_THREADS
+      if(!workerWakeupInited) {
+          pipe(workerWakeupPipe);
+          workerWakeupInited = rtsTrue;
+      }
+      FD_SET(workerWakeupPipe[0], &rfd);
+      maxfd = workerWakeupPipe[0] > maxfd ? workerWakeupPipe[0] : maxfd;
+#endif
+      
       /* Release the scheduler lock while we do the poll.
        * this means that someone might muck with the blocked_queue
        * while we do this, but it shouldn't matter:
@@ -137,135 +185,170 @@ awaitEvent(rtsBool wait)
        *
        * I believe none of these cases lead to trouble --SDM.
        */
+      
+#ifdef RTS_SUPPORTS_THREADS
+      isWorkerBlockedInAwaitEvent = rtsTrue;
+      workerWakeupPending = rtsFalse;
+#endif
       RELEASE_LOCK(&sched_mutex);
 
       /* Check for any interesting events */
-
-      tv.tv_sec = min / 1000000;
+      
+      tv.tv_sec  = min / 1000000;
       tv.tv_usec = min % 1000000;
 
-#ifndef linux_TARGET_OS
-      gettimeofday(&tv_before, (struct timezone *) NULL);
+      while ((numFound = select(maxfd+1, &rfd, &wfd, NULL, &tv)) < 0) {
+         if (errno != EINTR) {
+           /* Handle bad file descriptors by unblocking all the
+              waiting threads. Why? Because a thread might have been
+              a bit naughty and closed a file descriptor while another
+              was blocked waiting. This is less-than-good programming
+              practice, but having the RTS as a result fall over isn't
+              acceptable, so we simply unblock all the waiting threads
+              should we see a bad file descriptor & give the threads
+              a chance to clean up their act. 
+              
+              Note: assume here that threads becoming unblocked
+              will try to read/write the file descriptor before trying
+              to issue a threadWaitRead/threadWaitWrite again (==> an
+              IOError will result for the thread that's got the bad
+              file descriptor.) Hence, there's no danger of a bad
+              file descriptor being repeatedly select()'ed on, so
+              the RTS won't loop.
+           */
+           if ( errno == EBADF ) {
+             unblock_all = rtsTrue;
+             break;
+           } else {
+             fprintf(stderr,"%d\n", errno);
+             fflush(stderr);
+             perror("select");
+             barf("select failed");
+           }
+         }
+         ACQUIRE_LOCK(&sched_mutex);
+#ifdef RTS_SUPPORTS_THREADS
+          isWorkerBlockedInAwaitEvent = rtsFalse;
 #endif
 
-      while (!interrupted &&
-            (numFound = select(maxfd+1, &rfd, &wfd, NULL, &tv)) < 0) {
-       if (errno != EINTR) {
-         /* fflush(stdout); */
-         perror("select");
-         barf("select failed");
-       }
-       ACQUIRE_LOCK(&sched_mutex);
-
-       /* We got a signal; could be one of ours.  If so, we need
-        * to start up the signal handler straight away, otherwise
-        * we could block for a long time before the signal is
-        * serviced.
-        */
-       if (signals_pending()) {
-         RELEASE_LOCK(&sched_mutex);
-         start_signal_handlers();
-         break;
-       }
-
-       /* If new runnable threads have arrived, stop waiting for
-        * I/O and run them.
-        */
-       if (run_queue_hd != END_TSO_QUEUE) {
-         RELEASE_LOCK(&sched_mutex);
-         break;
-       }
-       
-       RELEASE_LOCK(&sched_mutex);
-      }        
-
-#ifdef linux_TARGET_OS
-      /* on Linux, tv is set to indicate the amount of time not
-       * slept, so we don't need to gettimeofday() to find out.
-       */
-      delta += min - (tv.tv_sec * 1000000 + tv.tv_usec);
-#else
-      gettimeofday(&tv_after, (struct timezone *) NULL);
-      delta += (tv_after.tv_sec - tv_before.tv_sec) * 1000000 +
-       tv_after.tv_usec - tv_before.tv_usec;
+         /* We got a signal; could be one of ours.  If so, we need
+          * to start up the signal handler straight away, otherwise
+          * we could block for a long time before the signal is
+          * serviced.
+          */
+#if defined(RTS_USER_SIGNALS)
+         if (signals_pending()) {
+             RELEASE_LOCK(&sched_mutex); /* ToDo: kill */
+             startSignalHandlers();
+             ACQUIRE_LOCK(&sched_mutex);
+             return; /* still hold the lock */
+         }
 #endif
 
-#if 0
-      if (delta != 0) { fprintf(stderr,"waited: %d %d %d\n", min, delta,
-                               interrupted); }
+         /* we were interrupted, return to the scheduler immediately.
+          */
+         if (interrupted) {
+             return; /* still hold the lock */
+         }
+         
+         /* check for threads that need waking up 
+          */
+         wakeUpSleepingThreads(getourtimeofday());
+         
+         /* If new runnable threads have arrived, stop waiting for
+          * I/O and run them.
+          */
+         if (run_queue_hd != END_TSO_QUEUE) {
+             return; /* still hold the lock */
+         }
+         
+#ifdef RTS_SUPPORTS_THREADS
+         /* If another worker thread wants to take over,
+          * return to the scheduler
+          */
+         if (needToYieldToReturningWorker()) {
+             return; /* still hold the lock */
+         }
 #endif
+         
+#ifdef RTS_SUPPORTS_THREADS
+          isWorkerBlockedInAwaitEvent = rtsTrue;
+#endif
+         RELEASE_LOCK(&sched_mutex);
+      }
 
       ACQUIRE_LOCK(&sched_mutex);
 
       /* Step through the waiting queue, unblocking every thread that now has
        * a file descriptor in a ready state.
-       
-       * For the delayed threads, decrement the number of microsecs
-       * we've been blocked for. Unblock the threads that have thusly expired.
        */
 
       prev = NULL;
-      for(tso = blocked_queue_hd; tso != END_TSO_QUEUE; tso = next) {
-       next = tso->link;
-       switch (tso->why_blocked) {
-       case BlockedOnRead:
-         ready = FD_ISSET(tso->block_info.fd, &rfd);
-         break;
-       
-       case BlockedOnWrite:
-         ready = FD_ISSET(tso->block_info.fd, &wfd);
-         break;
-       
-       case BlockedOnDelay:
-         {
-           int candidate; /* signed int is intentional */
-#if defined(HAVE_SETITIMER)
-           if (tso->block_info.delay > delta) {
-             tso->block_info.delay -= delta;
-             ready = 0;
-           } else {
-             tso->block_info.delay = 0;
-             ready = 1;
-           }
-#else
-           candidate = tso->block_info.target - getourtimeofday();
-           if (candidate < 0) {
-             candidate = 0;
-           }
-           if ((nat)candidate > delta) {
-             ready = 0;
-           } else {
-             ready = 1;
-           }
-#endif
-           break;
-         }
-       
-       default:
-         barf("awaitEvent");
-       }
+      if (select_succeeded || unblock_all) {
+         for(tso = blocked_queue_hd; tso != END_TSO_QUEUE; tso = next) {
+             next = tso->link;
+             switch (tso->why_blocked) {
+             case BlockedOnRead:
+                 ready = unblock_all || FD_ISSET(tso->block_info.fd, &rfd);
+                 break;
+             case BlockedOnWrite:
+                 ready = unblock_all || FD_ISSET(tso->block_info.fd, &wfd);
+                 break;
+             default:
+                 barf("awaitEvent");
+             }
       
-       if (ready) {
-         IF_DEBUG(scheduler,belch("Waking up thread %d\n", tso->id));
-         tso->why_blocked = NotBlocked;
-         tso->link = END_TSO_QUEUE;
-         PUSH_ON_RUN_QUEUE(tso);
-       } else {
+             if (ready) {
+                 IF_DEBUG(scheduler,belch("Waking up blocked thread %d\n", tso->id));
+                 tso->why_blocked = NotBlocked;
+                 tso->link = END_TSO_QUEUE;
+                 PUSH_ON_RUN_QUEUE(tso);
+             } else {
+                 if (prev == NULL)
+                     blocked_queue_hd = tso;
+                 else
+                     prev->link = tso;
+                 prev = tso;
+             }
+         }
+
          if (prev == NULL)
-           blocked_queue_hd = tso;
-         else
-           prev->link = tso;
-         prev = tso;
-       }
+             blocked_queue_hd = blocked_queue_tl = END_TSO_QUEUE;
+         else {
+             prev->link = END_TSO_QUEUE;
+             blocked_queue_tl = prev;
+         }
       }
-
-      if (prev == NULL)
-       blocked_queue_hd = blocked_queue_tl = END_TSO_QUEUE;
-      else {
-       prev->link = END_TSO_QUEUE;
-       blocked_queue_tl = prev;
+      
+#if defined(RTS_SUPPORTS_THREADS)
+       // if we were woken up by wakeBlockedWorkerThread,
+       // read the dummy byte from the pipe
+      if(select_succeeded && FD_ISSET(workerWakeupPipe[0], &rfd)) {
+          unsigned char dummy;
+          wait = rtsFalse;
+          read(workerWakeupPipe[0],&dummy,1);
       }
-
-    } while (wait && run_queue_hd == END_TSO_QUEUE);
 #endif
+    } while (wait && !interrupted && run_queue_hd == END_TSO_QUEUE);
+}
+
+
+#ifdef RTS_SUPPORTS_THREADS
+/* wakeBlockedWorkerThread
+ *
+ * If a worker thread is currently blocked within awaitEvent,
+ * wake it.
+ * Must be called with sched_mutex held.
+ */
+void
+wakeBlockedWorkerThread()
+{
+    if(isWorkerBlockedInAwaitEvent && !workerWakeupPending) {
+       unsigned char dummy = 42;       // Any value will do here
+       
+                       // write something so that select() wakes up
+       write(workerWakeupPipe[1],&dummy,1);
+       workerWakeupPending = rtsTrue;
+    }
 }
+#endif