(zilch statusbar): fix the deadlock

The statusbar code contained a deadlock in error cases, which, somehow,
was a heisenbug. After a 2-hour long debugging session today, it has
finally been spotted and _solved_.

I should some day move off Chicken Scheme.

Change-Id: If0c2a2b6db10b5620c12f191f13af4c86a6a6964

core/src/statusbar.sld

@@ -40,7 +40,22 @@
            (when start
              (redraw-status-bar))
            (mutex-unlock! mutex))))
-      (make-output-port (lambda (str) (mutex-lock! mutex) (write-data (string->utf8 str) 0)) close-this-port))
+      (make-output-port
+       (lambda (str)
+         ; In some rare cases, due to how Chicken Scheme handles errors,
+         ; it is possible for us to seemingly recursively have been called
+         ; while in the error handler. If this is the case (e.g. we are about
+         ; to attempt to lock a mutex we ostensibly own), bail out and unlock
+         ; the mutex.
+         (when (eq? (mutex-state mutex) (current-thread))
+           (mutex-unlock! mutex))
+
+         ; Try to force-disable the error handling behavior.
+         (##sys#print-length-limit #f)
+
+         (mutex-lock! mutex)
+         (write-data (string->utf8 str) 0))
+       close-this-port))
 
     ;; Creates a status bar. Ensures redraws are limited where necessary, and will erase itself before printing `stdout`. `stderr` output will be put in the statusbar.
     ;; if `print-logs` is `#t`, will output stderr to the display.