Re: sigprocmask improvement

[Bruno Haible]

Eric Blake wrote:
> found that it would somewhat easier if I could learn accurate
> signal information even when a signal is blocked.  Is this okay to apply?

Yes, of course. Thank you. From me, only a small comment update.


2008-06-22  Bruno Haible  <address@hidden>

        * lib/sigprocmask.c: Update a comment.

--- lib/sigprocmask.c.orig      2008-06-22 21:06:14.000000000 +0200
+++ lib/sigprocmask.c   2008-06-22 21:05:02.000000000 +0200
@@ -188,7 +188,8 @@
              {
                if (signal (sig, old_handlers[sig]) != blocked_handler)
                  /* The application changed a signal handler while the signal
-                    was blocked.  We don't support this.  */
+                    was blocked, bypassing our rpl_signal replacement.
+                    We don't support this.  */
                  abort ();
                received[sig] = pending_array[sig];
                blocked_set &= ~(1U << sig);
@@ -230,7 +231,8 @@
          old_handlers[sig] = handler;
          return result;
        }
-      return signal (sig, handler);
+      else
+       return signal (sig, handler);
     }
   else
     {