-
- /* Now, it goes without saying that the context sigmask
- * tweaking around this call is not pretty. However, it
- * currently seems to be "needed" for the following
- * situation. (So let's find a better solution and remove
- * this comment afterwards.)
- *
- * Suppose we are in a signal handler (let's say SIGALRM).
- * At the end of a WITHOUT-INTERRUPTS, the lisp code notices
- * that a thruption is pending, and says to itself "let's
- * receive pending interrupts then". We trust that the
- * caller is happy to run those sorts of things now,
- * including thruptions, otherwise it wouldn't have called
- * us. But that's the problem: Even though we can guess the
- * caller's intention, may_thrupt() would see that signals
- * are blocked in the signal context (because that context
- * itself points to a signal handler). So we cheat and
- * pretend that signals weren't blocked.
- * --DFL */
-#ifndef LISP_FEATURE_WIN32
- sigset_t old, *ctxset = os_context_sigmask_addr(context);
- unblock_signals(&deferrable_sigset, ctxset, &old);
-#endif