For the following test case (after JDK-8222717 is fixed) when a C2-compiled method calls a C1-compiled method, and the latter throws an exception, stack unwinding is broken and somehow we executed the "test47_value = 666" line, but with a wrong RSP so we get a crash there.

---------

    // C2->C1 invokestatic, make sure stack walking works (with static variable)
    @Test(compLevel = C2)
    public void test47(int n) {
        try {
            test47_helper(floatPointField, 1, 2, 3, 4, 5);
            test47_value = 666;
        } catch (MyRuntimeException e) {
            // expected;
        }
        test47_value = n;
    }

    @DontInline
    @ForceCompile(compLevel = C1)
    private static float test47_helper(FloatPoint fp, int a1, int a2, int a3, int a4, int a5) {
        test47_thrower();
        return 0.0f;
    }

    @DontInline @DontCompile
    private static void test47_thrower() {
        MyRuntimeException e = new MyRuntimeException("This exception should have been caught!");
        checkStackTrace(e, "test47_thrower", "test47_helper", "test47", "test47_verifier");
        throw e;
    }

    static int test47_value = 999;

    @DontCompile
    public void test47_verifier(boolean warmup) {
        int count = warmup ? 1 : 5;
        for (int i=0; i<count; i++) { // need a loop to test inline cache
            test47_value = 777 + i;
            test47(i);
            // FIXME JDK-8222908 Asserts.assertEQ(test47_value, i);
        }
    }