+ /* On Windows, there are no actual signals, but since the win32 port
+ * tracks the sigmask and checks it explicitly, some functions are
+ * still required to keep the mask set up properly. (After all, the
+ * goal of the sigmask emulation is to not have to change all the
+ * call sites in the first place.)
+ *
+ * However, this does not hold for all signals equally: While
+ * deferrables matter ("is interrupt-thread okay?"), it is not worth
+ * having to set up blockables properly (which include the
+ * non-existing GC signals).
+ *
+ * Yet, as the original comment explains it:
+ * Adjusting FREE-INTERRUPT-CONTEXT-INDEX* and other aspecs of
+ * fake_foreign_function_call machinery are sometimes useful here[...].
+ *
+ * So we merely skip this assertion.
+ * -- DFL, trying to expand on a comment by AK.
+ */