On 06/29/2011 06:24 PM, Yang, Wei wrote:
> +
> + /*
> + * Here we write the ro user page when
> + * cr0.wp=0, then we execute it and SMEP
> + * fault should happen.
> + */
> + err_prepare_notwp = ac_test_do_access(&at1);
> + if (!err_prepare_notwp) {
> + printf("%s: SMEP prepare fail\n", __FUNCTION__);
> + goto clean_up;
> + }
> +
> + at1.flags[AC_ACCESS_WRITE] = 0;
> + at1.flags[AC_ACCESS_FETCH] = 1;
> + ac_set_expected_status(&at1);
> + err_smep_notwp = ac_test_do_access(&at1);
> +
The address is accessed in the fist test, it is really "prefetch"-ed
in the second test?
>
> int ac_test_run(void)
> @@ -669,16 +765,22 @@ int ac_test_run(void)
> ac_test_t at;
> ac_pool_t pool;
> int i, tests, successes;
> + extern u64 ptl2[];
>
> printf("run\n");
> tests = successes = 0;
> ac_env_int(&pool);
> ac_test_init(&at, (void *)(0x123400000000 + 16 * smp_id()));
> do {
> + if (at.flags[AC_CPU_CR4_SMEP] && (ptl2[2] & 0x4))
> + ptl2[2] -= 0x4;
It seems you just remove the "U/S" bit, but forget to recover it, it can
break the test if AC_ACCESS_USER && !SMEP
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
