[Openmcl-devel] *multiprocessing-socket-io*

[R.Stoye]

since i missed to post gb's answer, i'll repost the rest of the  
conversation (per private mail) from 2008:

> If a thread is blocked in #_accept on a blocking socket in 1.1, I'm
> not sure that the thread can be killed by PROCESS-KILL.  (Driving
> a wooden stake through the heart of the thread may work ...)  This
> has to do with the fact that many blocking system calls are  
> automatically
> restarted when the signal used by PROCESS-INTERRUPT is received.
>
> The good news about that is that code which calls those blocking
> system calls doesn't have to check to see if an "interrupted system
> call" error (#$EINTR) occurred and manually retry the call.  The bad
> news is that when the handler for the interrupt signal is established
> with the SA_RESTART flag (as was the case in 1.1 and earlier), some
> system calls (including #_accept) will be automatically restarted
> and lisp won't regain control (and process the pending interrupt
> request) until #_accept returns "normally."
>
> When the socket's non-blocking, we don't generally block in
> #_accept (but instead wait interruptibly for input to be available);
> if PROCESS-KILL calls PROCESS-INTERRUPT to tell the thread to
> kill itself, the interruptible wait completes and as it's returning
> to lisp it handles the interrupt request and starts to kill itself.
>
> Just to see if I'm on the right track here, could you check to
> see if the thread is still around (quite possibly still blocked
> in #_accept) when the listener socket is blocking and PROCESS-KILL
> thinks that it's just killed the blocked thread ?
>
> If this (#_accept not getting interrupted) is the root of the
> problem, then the best thing to do in the short term is to make the  
> socket non-blocking (by leaving *multiprocessing-socket-io*
> set to T).  1.2 doesn't make blocking system calls automatically
> restartable and is supposed to check each such system call for
> #$EINTR and manually restart it when that value is returned.


Am 10.03.2008 um 16:03 schrieb Gary Byers:

> Just to see if I'm on the right track here, could you check to
> see if the thread is still around (quite possibly still blocked
> in #_accept) when the listener socket is blocking and PROCESS-KILL
> thinks that it's just killed the blocked thread ?

yes, the thread is "still around":
the kill "proceeds" when a connection arives.
so i stay with *multiprocessing-socket-io* set to T

;; test:

;; simplest server waiting on 9001
(defun runaboo ()
  (let ((theprocess (make-process "runaboo")))
    (flet ((co-handler (client-stream)
	     (format T "~&Connection handler...")
	     (unwind-protect (print (read-line client-stream))
	       (close client-stream :abort t) )))
      (process-preset theprocess
		      #'(lambda ()
			  (let ((server-socket (ccl:make-socket
						:connect :passive
							 :local-port 9001
							 :reuse-address t
							 :external-format (make-external-format :character- 
encoding :iso-8859-1 :line-termination :crlf)
							 ))
				(client-streams nil))
			    (unwind-protect
				 (progn (format t "~%entered unwind~%")
					(loop
					   (format t "~%entered loop waiting for client~%")
					   (let* ((client-stream (ccl:accept-connection server- 
socket :wait t)))
					     (push client-stream client-streams)
					     (format t "~%have client on stream ~s~%" client-stream)
					     (describe client-stream)
					     (unless client-streams
					       (error "no client???"))
					     (funcall #'co-handler client-stream)
					     (setf client-streams (remove client-stream client-streams) )))
					(format t "~%after loop~%"))
			      (format t "~%closing server~%")
			      (format t "the clients are: ~s~%" client-streams)
			      (mapcar #'(lambda (s) (ignore-errors (close s :abort t)))  
client-streams)
			      (format t "~%close server~%")
			      (close server-socket :abort t)
			      )))))
    (process-enable theprocess)
    theprocess))

? (setf ccl::*multiprocessing-socket-io* nil)

? (setf myboo (runaboo))
#<PROCESS runaboo(4) [Active] #x300041072C5D>
?
entered unwind

entered loop waiting for client
(ccl::process-kill myboo)
T
? (room t)
Approximately 29,753,344 bytes of memory can be allocated
before the next full GC is triggered.

                   Total Size             Free                 Used
Lisp Heap:       47448064 (46336K)   29753344 (29056K)   17694720  
(17280K)
Stacks:          18876320 (18434K)   18870640 (18428K)       5680 (6K)
Static:           1530144 (1494K)           0 (0K)        1530144  
(1494K)
506834.750 MB reserved for heap expansion.

Initial(0)
  cstack:      2504608 (2446K)      2504608 (2446K)            0 (0K)
  vstack:      2371584 (2316K)      2371296 (2316K)          288 (0K)
  tstack:       327680 (320K)        326928 (319K)           752 (1K)
listener(1)
  cstack:      2506752 (2448K)      2506752 (2448K)            0 (0K)
  vstack:      2166784 (2116K)      2165144 (2114K)         1640 (2K)
  tstack:      2162688 (2112K)      2159632 (2109K)         3056 (3K)
runaboo(4)
  cstack:      2506752 (2448K)      2506752 (2448K)            0 (0K)
  vstack:      2166784 (2116K)      2166776 (2116K)            8 (0K)
  tstack:      2162688 (2112K)      2162688 (2112K)            0 (0K)
NIL
?

;; in another Terminalwindow:
ps axMlwww | grep 6844
rs        6844 s006    0.0 S     31   0:00.16   0:00.20   501  6789    
0  0  2715132  19176 -        0:00.99 /usr/local/src/ccl/dx86cl64
          6844         0.0 S     31   0:00.01   0:00.00   501  6789    
0  0  2715132  19176 -        0:00.99
          6844         0.0 S     31   0:00.08   0:00.53   501  6789    
0  0  2715132  19176 -        0:00.99
          6844         0.0 S     31   0:00.00   0:00.00   501  6789    
0  0  2715132  19176 -        0:00.99

;; so the runaboo really exists!
;; so does the listening socket:
[rs at macbook:~]$ netstat -an | grep 9001
tcp4       0      0  *.9001                 *.*                     
LISTEN

;; now connect per telnet
telnet localhost 9001
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.

;; meanwhile in openmcl:
?
closing server
the clients are: NIL

close server

;;; and after that the runaboo thread is gone, netstat reports nothing  
on 9001