x86/mpx, selftests: Only check bounds-vs-shadow when we keep shadow

The check between the hardware state and our shadow of it is
checked in the signal handler for all bounds exceptions,
even for the ones where we don't keep the shadow up2date.
This is a problem because when no shadow is kept the handler
fails at this point and hides the real reason of the
exception.

Move the check into the code-path evaluating normal bounds
exceptions to prevent this.

Signed-off-by: Joerg Roedel <jroedel@suse.de>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kselftest@vger.kernel.org
Link: http://lkml.kernel.org/r/1491488598-27346-1-git-send-email-joro@8bytes.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>

tools/testing/selftests/x86/mpx-mini-test.c

@@ -404,8 +404,6 @@ void handler(int signum, siginfo_t *si, void *vucontext)
 		dprintf2("info->si_lower: %p\n", __si_bounds_lower(si));
 		dprintf2("info->si_upper: %p\n", __si_bounds_upper(si));
 
-		check_siginfo_vs_shadow(si);
-
 		for (i = 0; i < 8; i++)
 			dprintf3("[%d]: %p\n", i, si_addr_ptr[i]);
 		switch (br_reason) {
@@ -416,6 +414,9 @@ void handler(int signum, siginfo_t *si, void *vucontext)
 			exit(5);
 		case 1: /* #BR MPX bounds exception */
 			/* these are normal and we expect to see them */
+
+			check_siginfo_vs_shadow(si);
+
 			dprintf1("bounds exception (normal): status 0x%jx at %p si_addr: %p\n",
 				status, (void *)ip, si->si_addr);
 			num_bnd_chk++;