gmain: Fix use of atomic primitives with sig_atomic_t

It seems that `sig_atomic_t` is not the same width as `int` on FreeBSD,
which is causing CI failures:
```
 ../glib/gmain.c:5206:3: error: '_GStaticAssertCompileTimeAssertion_73' declared as an array with a negative size
  g_atomic_int_set (&any_unix_signal_pending, 0);
  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../glib/gatomic.h:100:5: note: expanded from macro 'g_atomic_int_set'
    G_STATIC_ASSERT (sizeof *(atomic) == sizeof (gint));                     \
    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
```

Fix that by only using `sig_atomic_t` if the code is *not* using atomic
primitives (i.e. in the fallback case). `sig_atomic_t` is only a typedef
around an integer type and is not magic. Its typedef is chosen by the
platform to be async-signal-safe (i.e. read or written in one instruction),
but not necessarily thread-safe.
Signed-off-by: Philip Withnall <withnall@endlessm.com>

glib/gmain.c

@@ -452,13 +452,22 @@ static GMainContext *glib_worker_context;
  * Both variables must be accessed using atomic primitives, unless those atomic
  * primitives are implemented using fallback mutexes (as those aren’t safe in
  * an interrupt context).
+ *
+ * If using atomic primitives, the variables must be of type `int` (so they’re
+ * the right size for the atomic primitives). Otherwise, use `sig_atomic_t` if
+ * it’s available, which is guaranteed to be async-signal-safe (but it’s *not*
+ * guaranteed to be thread-safe, which is why we use atomic primitives if
+ * possible).
+ *
+ * Typically, `sig_atomic_t` is a typedef to `int`, but that’s not the case on
+ * FreeBSD, so we can’t use it unconditionally if it’s defined.
  */
-#ifdef HAVE_SIG_ATOMIC_T
-static volatile sig_atomic_t unix_signal_pending[NSIG];
-static volatile sig_atomic_t any_unix_signal_pending;
-#else
+#if (defined(G_ATOMIC_LOCK_FREE) && defined(__GCC_HAVE_SYNC_COMPARE_AND_SWAP_4)) || !defined(HAVE_SIG_ATOMIC_T)
 static volatile int unix_signal_pending[NSIG];
 static volatile int any_unix_signal_pending;
+#else
+static volatile sig_atomic_t unix_signal_pending[NSIG];
+static volatile sig_atomic_t any_unix_signal_pending;
 #endif
 
 /* Guards all the data below */

[Philip Withnall]

I accidentally pushed this straight to master rather than a feature branch on origin. :(

@sdroege, would you mind doing a post-hoc review? If there are any issues I can revert this and put it through an MR.

[Sebastian Dröge]

Seems fine to me

[Philip Withnall]

Thanks, sorry for the cock-up.