Fix an off-by-one error in the sigset_t function error handling.

Spotted while running the tests on MIPS, where sigset_t is
actually large enough. The bits in sigset_t are used such that
signal 1 is represented by bit 0, so the range of signals is
actually [1, 8*sizeof(sigset_t)]; it seems clearer to reword
the code in terms of valid bit offsets [0, 8*sizeof(sigset_t)),
which leads to the usual bounds checking idiom.

Change-Id: Id899c288e15ff71c85dd2fd33c47f8e97aa1956f

tests/signal_test.cpp

@@ -48,13 +48,13 @@ static void TestSigSet2(Fn fn) {
   int min_signal = SIGHUP;
   int max_signal = SIGRTMAX;
 
-#if __BIONIC__
-  // bionic's sigset_t is too small: 32 bits instead of 64.
+#if defined(__BIONIC__) && !defined(__mips__)
+  // bionic's sigset_t is too small for ARM and x86: 32 bits instead of 64.
   // This means you can't refer to any of the real-time signals.
   // See http://b/3038348 and http://b/5828899.
-  max_signal = 31;
+  max_signal = 32;
 #else
-  // Other C libraries are perfectly capable of using their largest signal.
+  // Other C libraries (or bionic for MIPS) are perfectly capable of using their largest signal.
   ASSERT_GE(sizeof(sigset_t) * 8, static_cast<size_t>(SIGRTMAX));
 #endif