Kill the administrative threads on error.

scsh/procobj.scm

@@ -172,14 +172,13 @@
 
 (define (with-autoreaping thunk)
   (set! *autoreap-policy* 'early)
-  (spawn (lambda ()
-	   (let lp ((event (most-recent-sigevent)))
-	     (let ((next-event (next-sigevent event interrupt/chld)))
-	       (*sigchld-handler*)
-	       (lp next-event))))
-	 '*sigchld-handler*-thread)
-  (thunk))
-
+  (run-as-long-as
+   (lambda ()
+     (let lp ((event (most-recent-sigevent)))
+       (let ((next-event (next-sigevent event interrupt/chld)))
+	 (*sigchld-handler*)
+	 (lp next-event))))
+   thunk))
 
 ;;; This list contains pids whose proc-obj were gc'd before they died
 ;;; We try to reap them after every gc and maybe on every SIGCHLD

scsh/sighandlers.scm

@@ -188,9 +188,7 @@
 				   (structure-ref threads-internal event-type) 
 				   interrupt)
 				  (enum interrupt keyboard))))))
-  (spawn deliver-interrupts
-	 'deliver-interrupts)
-  (thunk))
+  (run-as-long-as deliver-interrupts thunk))
 
 (define (deliver-interrupts)
   (let lp ((last ((structure-ref scsh-events most-recent-sigevent))))

scsh/utilities.scm

@@ -267,3 +267,24 @@
 (define-record-resumer :reinitializer
   (lambda (r)
     ((reinitializer-thunk r))))
+
+;--------------
+; Run thunk1 until thunk2 escapes
+; This is *extremly* low level
+; Don't use unless you know what you are doing
+
+(define (run-as-long-as thunk1 thunk2)
+  (let ((thread (make-placeholder)))
+    (spawn (lambda ()
+	     (placeholder-set! thread (current-thread))
+	     (thunk1)))
+    (dynamic-wind
+     (lambda () #t)
+     thunk2
+     (lambda ()
+       (remove-thread-from-queues! (placeholder-value thread))
+       (kill-thread! (placeholder-value thread))
+       (make-ready (placeholder-value thread))))))
+
+    
+