[Chicken-users] Re: clueless about memory consumption

[F. Wittenberger]

Hi,

while I'm still not sure that I found the issue, at least I on spotted
something, which is definitely wrong in chicken: mutex-unlock!

Index: srfi-18.scm
===================================================================
--- srfi-18.scm
+++ srfi-18.scm
@@ -368,11 +368,12 @@
        (lambda (return)
         (let ([waiting (##sys#slot mutex 3)]
               [limit (and timeout (##sys#compute-time-limit
timeout))] )
           (##sys#setislot mutex 4 #f)
           (##sys#setislot mutex 5 #f)
-          (##sys#setslot ct 8 (##sys#delq mutex (##sys#slot ct 8)))
+          (let ((t (##sys#slot mutex 2)))
+            (##sys#setslot t 8 (##sys#delq mutex (##sys#slot t 8))))
           (when cvar
                 (##sys#setslot cvar 2 (##sys#append (##sys#slot cvar 2)
(##sys#list ct)))
                 (##sys#setslot ct 11 cvar)
                 (cond [limit
                        (##sys#setslot

Upon unlock currently chickens code will *try* to remove the mutex from
those held by the *current* thread.  This works under the theory that
the mutex has been locked by the thread which unlocks it.  This might
not be the case!  It's legal unser srfi-18 to unlock a mutex locked by
another thread and it's a well known pattern in thread communication.

This can have two spurious effects: a) unexpected
abandoned-mutex-exceptions for properly unlocked mutex b) when the
thread, which locked the mutex runs long enough, it will collect a huge
pile of owned, but unlocked mutex.  (The latter could be my issue, we'll
see).


Thanks for your help

/Jörg